You are on page 1of 9

Journal of Experimental Psychology: Copyright 2005 by the American Psychological Association

Human Perception and Performance 0096-1523/05/$12.00 DOI: 10.1037/0096-1523.31.1.101


2005, Vol. 31, No. 1, 101–109

Psychophysical Isolation of the Modality Responsible for Detecting


Multimodal Stimuli: A Chemosensory Example
Hisanori Nagata, Pamela Dalton, Nadine Doolittle, and Paul A. S. Breslin
Monell Chemical Senses Center

Multiple sense modalities can be stimulated conjointly by a physically complex item, such as a predator,
and also by a physically solitary stimulus that acts on multiple receptor classes. As a prime example of
this latter group, l-menthol from mint stimulates taste, smell, and several somatosensory submodalities.
In 6 experiments that used a variety of psychometric techniques, the authors experimentally isolated the
modality by which l-menthol is detected in the upper airways (the nose and mouth). Interestingly,
absolute detection in both the nasal and oral cavities was based on olfaction and not stinging, cooling,
or taste. These experiments illustrate how the sensory modality responsible for detecting a multimodal
or multisensory stimulus can be psychophysically isolated.

The ability to simultaneously use multiple sensory inputs to (cooling, irritation, tingling, etc.), olfaction (minty), and gustation
perceive the external environment is an adaptive characteristic of (bitter). Regarding menthol, complex sensory and perceptual in-
most vertebrates. Many real world entities, such as foods, preda- teractions take place within the chemesthetic modality (Cliff &
tors, and even conspecific animals, are physically complex stim- Green, 1993, 1996; Green, 1985), from the activation of submo-
uli— emitting sounds, reflecting light, and releasing chemicals into dalities such as irritation (burning, prickling, stinging, itching, etc.)
the air. Physically simple stimuli, such as chemicals, may also to thermal effects (cooling, warming) and tactile effects (buzzing,
stimulate multiple sensory systems and subsystems. To wit, recent tingling, tickling, numbing, etc.). In addition to stimulating odor
evidence within the chemical senses suggests that traditional and taste, menthol elicits stinging, numbing, touch, and a cooling
bitter-tasting compounds activate somatosensory neurons (Finger sensation, which can inhibit the perception of warmth (Green,
et al., 2003; Liu & Simon, 1998), thermal stimuli induce taste 1986, 1992).
sensations (Cruz & Green, 2000), astringent compounds (alum) Humans experience l-menthol most commonly in the upper
elicit taste sensations (Breslin, Gilmore, Beauchamp, & Green, airways in the form of lozenges, chewing gums, cigarettes, tooth-
1993), and a large majority of olfactory stimuli elicit irritation in pastes, and mouth washes. Humans can detect the presence of
the nose (Cometto-Muñiz & Cain, 1995) and possibly pharynx. l-menthol in both the nasal and oral cavities; however, the relative
Under these circumstances, it is very difficult to know which sense sensitivity of each of these regions and the modality by which
modality is responsible for detection of the stimulus or object, absolute detection of menthol occurs has not been well established.
especially when qualitative information is lacking, as is often the Thus, as a prototypical example of a compound that stimulates
case at absolute detection (Stevens, 1995, 1997). There are even multiple sensory modalities, menthol could be detected using any
individual stimuli that will stimulate several sensory systems si- of these sensory systems in either the nose or mouth.
multaneously, including, taste, smell, pain, tactile, and thermal To illustrate this point, we conducted six experiments to psy-
sensations. The principal flavor in mint, l-menthol, is a prototyp- chophysically isolate the sensory modality responsible for the
ical example of this type of stimulus. absolute detection of l-menthol. In Experiment 1, absolute detec-
That a single physical stimulus may have multimodal action is tion thresholds were measured in the nasal cavity for l-menthol in
a widely accepted idea. To this point, l-menthol is widely used as polyethylene glycol (PEG) 200. Oral detection thresholds with
both a chemical cooling agent and as a counterirritant, that is, an aqueous solutions were collected with and without nose clips to
analgesic, for intrinsic irritation (e.g., from oral and pharyngeal determine the effect of retronasal olfaction on oral thresholds
epithelia and from muscle). In the upper airways, l-menthol can (Experiments 2, 3A, and 3B). Volatile compounds do not flow
elicit sensations using three different modalities: somatosensation retronasally into the nasal cavity when the nares are pinched close.
Thus, when one’s nares are pinched in an oral detection threshold
paradigm, oral sensations can be effectively isolated from nasal
Hisanori Nagata, Pamela Dalton, Nadine Doolittle, and Paul A. S. sensations. Because intranasal chemosensation can be based on
Breslin, Monell Chemical Senses Center, Philadelphia, Pennsylvania. either olfactory or somatosensory detection, we took advantage of
This research was supported by National Institutes of Health (NIH) the fact that an olfactory sensation cannot be localized, or even
Grant RO1DC03704 to Pamela Dalton and NIH Grant DC02995 to Paul
lateralized in humans (Kobal, Van Toller, & Hummel, 1989), and
A. S. Breslin. We thank Ines Rodriguez and Jeanmarie Diamond for their
collected intranasal lateralization thresholds to identify whether
assistance with laboratory work and Barry Green for his advice regarding
olfactory lateralization techniques. nasal detection is olfactory or somatosensory based (Experiments
Correspondence concerning this article should be addressed to Paul 4 and 5). Experiments 3A, 3B, 4, and 5 attempted to refine the
A. S. Breslin, Monell Chemical Senses Center, 3500 Market Street, Phil- possible modalities responsible for the oral and nasal detection of
adelphia, PA 19104. E-mail: breslin@monell.org l-menthol. Specifically, we asked whether oral thresholds were

101
102 NAGATA, DALTON, DOOLITTLE, AND BRESLIN

based on the taste, the odor, or the chemesthetic sensations in the oral (in deionized Millipore filtered water) stimuli. This was to determine
oral cavity from l-menthol (Experiments 3A and 3B), and once whether the concentrations of volatile l-menthol were similar for the two
oral detection was determined to be actually retronasal in origin, methods and thereby link them using a common physical basis for detec-
we asked whether nasal thresholds were based on odor or chem- tion, which in turn implicates the modality of detection. Vapor concentra-
tion of l-menthol in both solvents was measured using a Hewlett-Packard
esthetic sensations in the nasal passages (Experiments 4 and 5).
6890 gas chromatograph equipped with a flame ionization detector linked
with a Hewlett-Packard Autosampler (HP7694). Splitless samples were
General Experimental Design injected onto a SUPELCOWAX 10 capillary column (0.53-mm inner
diameter, 30-m long, film thickness ⫽ 0.50 ␮m) at a temperature program
Participants that ramped from 100° C (1 min) to 180° C (10° C/min) to 220° C (25°
C/min), and detection occurred at 250° C. A four-point calibration curve
All participants were paid volunteers who provided signed consent using
for l-menthol was generated by injecting known amounts of l-menthol
a form that was approved by the Office of Regulatory Affairs at the
(10.0, 31.6, 100.0, and 316.0 ppm); their associated chromatographic areas
University of Pennsylvania. Participants were instructed not to eat, drink,
were used to convert the areas into concentration units (ppm ⫻ volume).
chew gum, or smoke for at least 1 hr prior to testing.
The coefficient of determination (r2) for the calibration curve was greater
than .98.
Procedures: Absolute Detection Thresholds (Experiments
1, 2, 3, and 5)
Experiment 1: Nasal Detection Thresholds
Absolute detection thresholds were measured with the two alternative
forced-choice method described by Stevens (1995). On each trial, the To our knowledge, there are no studies directly comparing the
participant was presented with both a stimulus and a blank in a random sensitivity to l-menthol in the nose and in the mouth. Therefore,
order. After sampling the two stimuli (cups or bottle pairs) sequentially, the baseline data must be collected to determine the absolute detection
participant was asked to identify which sample (first or second) contained thresholds for menthol in the two main upper airway passages.
the stimulus. An incorrect response on any trial resulted in presentation of Experiment 1 measured nasal detection thresholds in 100 volun-
the next higher concentration, whereas four consecutive correct responses teers in duplicate, and Experiment 2 measured oral detection
at a given concentration resulted in the presentation of the next lower thresholds to menthol in triplicate. Experiment 1 involved an order
concentration (four-down, one-up staircase criteria). Testing continued
of magnitude 10 times more participants than did Experiment 2
until five reversals (a correct decision followed by an incorrect decision or
because it also generated data to help determine whether the
vice versa) were collected. If, at any time, a participant moved more than
two consecutive concentration steps either up or down from the concen- frequency histogram for sensitivity to l-menthol would deviate
tration at which the first reversal occurred, then all reversals up to that point from unimodality, possibly indicating a small number of mecha-
were disregarded and five more reversals were required. These constraints nisms involved in detection, as is seen with distributions that
were used in order to clamp the acceptable concentration range of reversals appear bimodal.
to within an order of magnitude. If the participant achieved five reversals
but did not meet the additional criteria, testing was continued until the
criteria were met. The threshold was calculated from the mean of the last
Method
four reversals. Testing could be terminated under four different conditions: Participants. One hundred volunteers were paid to participate after
(a) five valid reversals were achieved and the extra criteria were met, signing an Institutional Review Board approved consent form (as was true
resulting in a threshold value; (b) a participant failed to correctly detect the in all subsequent experiments below). Participants ranged in age from 16
stimulus at the highest concentration; (c) the participant correctly identified to 69 years (average age ⫽ 33.5 years), with equal representation of males
the stimulus on four consecutive trials at the lowest concentration; (d) the and females. Fifteen participants had previous experience in psychophys-
participant did not meet the criteria for a threshold before the 40th trial, ical experiments.
whereupon the session was terminated to prevent fatigue. Participants were Stimuli. The nasal series included 26 samples ranging in concentration
tested without trial feedback to focus their attention on their chemosensa- from 1.56 ⫻10⫺9 M to 0.1016 M l-menthol in PEG200 in one quarter log
tions, rather than teach them to attend to any correlated extraneous cues steps. A total of six blanks were used (PEG200 only), and new blanks were
(which we attempted to minimize; see below for details). prepared daily. Nasal threshold series were made on Monday and used for
1 week only. Stimuli were presented to participants in 280-ml glass bottles
Data Analysis fitted with specially designed Teflon nose pieces. Glass bottles contained
either 10 ml of a solution of l-menthol in PEG200 solution (odorant) or the
Preliminary evaluation of the data indicated that the threshold ranges same amount of PEG200 (blank). The bottles were wrapped in thin foam
were distributed lognormally. Consequently, the threshold concentration material to help eliminate any external temperature influences (bottles were
values were transformed to logarithms before statistical analysis. Test– maintained at a room temperature of approximately 21° C) and to thermally
retest reliabilities for nasal detection thresholds and oral detection thresh- insulate the bottles when handled.
olds were assessed by Pearson’s product–moment correlation coefficients Procedure. Participants and experimenters handled bottles while wear-
and two-tailed t tests. The difference between the oral detection thresholds ing polypropylene gloves to further reduce transmission of extraneous cues
with and without nose clips was evaluated using a t test for paired samples. in the detection task. On each trial, the participant was presented with two
A p value was selected for two-tailed significance. For all tests of signif- pairs of glass bottles. One pair contained l-menthol and a blank (odorant
icance, the alpha error level was set at p ⫽ .05. All data analysis was pair); the other pair contained two blanks only (blank pair). The partici-
analyzed by SPSS software (SPSS, Inc., Chicago, IL). pants simultaneously sniffed from both bottles within a pair and sniffed
from both pairs sequentially, then decided which pair of bottles (first or
Gas Chromatography second) contained the odorant. The l-menthol was placed randomly on the
left or right bottle within the odorant pair and was presented randomly
We used headspace analysis using gas chromatography to quantify the either in the first or second set of bottles. To avoid adaptation and/or
amount of l-menthol in the headspace of both the nasal (in PEG200) and sensitization, we used a 30-s intertrial interval (ITI). We selected this
PSYCHOPHYSICAL ISOLATION OF SENSORY MODALITY 103

interval because pilot testing indicated that 30 to 45 s minimized both


sensitization and desensitization to irritation over several suprathreshold
l-menthol exposures in 6 participants. The pilot test requested that they
sample 10 menthol solutions at a fixed intersolution interval ranging from
10 s to 5 min and rate irritation from l-menthol after each sampling.
Different ITIs were tested on different days. The 30-s ITI was extended
slightly if a participant required more than 30 s to complete a trial. The test
began with Bottle Number 3 (0.0065 M). Each participant completed two
nasal detection threshold sessions with a 15-min break between the two
sessions. The participant left the test room during the break.

Results and Discussion


Individual threshold concentrations varied from 5.00 ⫻ 10⫺5 M
to 5.10 ⫻ 10⫺2 M (three orders of magnitude variation) l-menthol Figure 2. The test–retest reliability of l-menthol nasal detection thresh-
in PEG200 (relative headspace concentration ranging from 0.002 olds for 100 participants in Experiment 1. The x-axis represents the
␮g/L to 11.600 ␮g/L), with an exception of two data points threshold value of each participant from Test 1 in log molar. The y-axis
(3.00 ⫻ 10⫺3 M, 1.20 ⫻ 10⫺2 M), and with an overall geometric represents the threshold value of each participant from Test 2 in log molar.
mean threshold concentration of 3.42 mM (see Figure 1). The The test–retest reliability was highly significant. The slope departs from
frequency distribution of nasal thresholds was unimodal, appeared unity because a few participants had much lower thresholds on Test 2.
approximately normal, and was found to be more consistent and to
fall within a tighter range than has been previously reported for the 23.5 years). All participants had previous experience in psychophysical
nasal thresholds of many other olfactory compounds. Interestingly, experiments.
among the 100 participants tested, there were no significant rela- Stimuli. The oral threshold series contained 13 samples, ranging in
tionships between nasal sensitivity to l-menthol and age, gender, or concentration from 4.80 ⫻ 10⫺7 M to 4.75 ⫻ 10⫺4 M l-menthol in
smoking status. Test–retest reliability was quite good (r ⫽ .79, p ⬍ deionized Millipore filtered water in one quarter log steps. Deionized
.01; see Figure 2). The slope deviates slightly from unity because Millipore filtered water was used for blanks as well as for rinsing between
a few participants had much lower thresholds on their second test. samples. All stimuli and blanks were maintained at a constant temperature
by immersion of bottles in a circulating water bath (Cole-Parmer Polystat)
held at 25° C. Fresh oral threshold series were prepared weekly.
Experiment 2: Oral Detection Thresholds With the Nose Procedure. On each trial, participants were given two samples (10 ml
Open each) in 40-ml plastic medicine cups (Baxter). They were instructed to take
the first sample, hold it in their mouth for 5 s, expectorate, and rinse.
Method Participants then repeated this procedure with the second sample and
Participants. Ten volunteers (4 women and 6 men) were paid to decided which sample contained the stimulus. Participants breathed natu-
participate. Participants ranged in age from 18 to 32 years (average age ⫽ rally through the nose while resting between trials. Cups were presented in
random order. A 45-s ITI was used for oral thresholds to minimize
sensitization and desensitization effects. Three thresholds were collected
on different days.

Results and Discussion


Averaged oral threshold concentration for each participant
ranged from 1.40 ⫻10⫺6 M to 1.43 ⫻10⫺4 M l-menthol in water
with an overall geometric mean threshold concentration of 9.00 ⫻
10⫺6 M (see Figure 3). The distribution of oral detection thresh-
olds was also unimodal and spanned more than two orders of
magnitude. Test–retest reliability was good and increased slightly
between Test 1-Test 2 (r ⫽ .61, p ⬍ .01) and Test 2-Test 3 (r ⫽
.67, p ⫽ .17; see Figure 4).

Experiment 3A: Oral Detection Thresholds With the Nose


Closed or Open
Experiments 1 and 2 allow one to compare nasal and oral
absolute detection thresholds with l-menthol and revealed certain
Figure 1. A frequency histogram of 200 l-menthol nasal detection thresh-
similarities. For example, both frequency-histogram distributions
olds (two thresholds from 100 participants) from Experiment 1. The y-axis
depicts the number of participants in each threshold category bin. The
were unimodal and spanned approximately two to three orders of
x-axis depicts the various millimolar concentration bin categories of magnitude (compare Figures 1 and 3). These data do not, however,
l-menthol in polyethylene glycol (PEG) 200. The distribution is approxi- indicate the sensory modality by which menthol was detected.
mately lognormal, spanning three orders of magnitude, and is unimodal in Thus, oral menthol detection could be based on taste, odor, or
form. The geometric mean nasal threshold level was 3.42 mM. somatosensory sensations. Experiment 3 compared oral threshold
104 NAGATA, DALTON, DOOLITTLE, AND BRESLIN

testing when the nares were open or when they had been closed
with nose clips. There can be no retronasal flow of volatile
compounds to the nasal cavity from the mouth when the nares are
pinched closed. Therefore, this simple technique serves to isolate
the two upper respiratory passageways. If the detection thresholds
are the same, then detection of menthol is oral and olfaction can be
ruled out. If the oral detection threshold is higher when the nares
are closed, then retronasal flow of volatiles will appear critical for
oral detection.

Method
Participants. Participants were the same as in Experiment 2.
Stimuli. Samples were the same as in Experiment 2.
Procedure. To determine the effect of retronasal olfaction on the oral
threshold for l-menthol, we compared the oral threshold for l-menthol in
Figure 3. A frequency histogram of 29 l-menthol oral detection thresh- water when the retronasal flow of l-menthol vapor was permitted and the
olds (three thresholds from 10 participants; 1 participant was unable to test oral threshold for l-menthol when retronasal flow was blocked by wearing
a third time) from Experiment 2. The y-axis depicts the number of partic- plastic and foam rubber nose clips (Diagnostic Medical System). In the first
ipants in each threshold category bin. The x-axis depicts the various condition, the participants were instructed to put on the nose clips, rinse,
micromolar concentration bin categories of l-menthol in deionized Milli- take the first sample, hold for 5 s, and then expectorate. The participants
pore filtered water. The distribution is approximately unimodal in form and would follow the same procedure with the second sample, followed by a
spans more than two orders of magnitude. The geometric mean oral judgment of which sample contained the stimulus. The participants could
threshold level was 9.0 ⫻10⫺6 M. then remove the nose clips until the next set of samples was presented.
Participants were not instructed how to breathe during the ITI.
In the second condition, thresholds were obtained without nose clips.
Participants were instructed to rinse, take the first sample, hold for 5 s

Figure 4. The test–retest reliability of l-menthol oral detection thresholds for the 9 participants in Experiment
2 who completed all three oral thresholds. All axes depict the oral threshold concentration in log millimolar
l-menthol. Test 1 to Test 2 reliability is depicted in the top panel, Test 2 to Test 3 reliability is in the lower left
panel, and Test 1 to Test 3 reliability is in the lower right panel.
PSYCHOPHYSICAL ISOLATION OF SENSORY MODALITY 105

while breathing through the nose, and then expectorate. After following the Table 2
same procedure for the second sample, they were asked to judge which cup Individual Geometric Means of Oral Threshold With and
contained the stimulus. Because the purpose of this study was to determine Without Nose Clips in Experiment 3B
the role of retronasal olfaction on the detection of an orally presented
stimulus, we wanted to maintain retronasal olfaction as sensitive as pos- Participant mM
sible. Therefore, during the ITI, participants were instructed to breathe only
through the mouth to avoid any possible adaptation of olfaction and/or Without nose clips
pharyngeal irritation from residual l-menthol vapor in the upper airways or 1 0.1240
in the ambient room air. Participants were tested in a total of three sessions. 2 0.1182
3 0.0604
Within a single session, each participant was tested in both conditions with
4 0.0089
a 15-min break between conditions. Order of conditions was counterbal- 5 0.1126
anced across sessions and, as always, stimulus order was randomized. 6 0.0604
7 0.1576
Results and Discussion 8 0.1909
9 0.0130
The results of Experiment 3A are shown in Table 1; the results 10 0.0523
of Experiment 3B (discussed below) are shown in Table 2. Thresh- Geometric M 0.0636
olds for oral l-menthol obtained without retronasal flow (wearing With nose clips
nose clips) were significantly greater ( p ⬍ .01), by an order of
magnitude, than thresholds for oral l-menthol obtained when ret- 1 0.2205
ronasal flow was unimpeded (7.46 ⫻ 10⫺5 M vs. 7.70 ⫻ 10⫺6 M; 2 0.2546
3 0.2426
see Table 1). These oral thresholds both with and without nose 4 0.0499
clips show little relationship (r2 ⫽ .11; negative trend) and, in light 5 0.2802
of our high test–retest reliability (see Figures 2 and 4), suggest that 6 0.1126
these thresholds are based on independent mechanisms. 7 0.0767
The headspace concentration of samples used in Experiments 1, 8 0.2940
9 0.0731
2, and 3 were analyzed with gas chromatography. Figure 5 shows 10 0.0548
Geometric M 0.1345

Table 1 Note. This table presents the results of Experiment 3B and parallels Table
Individual Geometric Means of Oral Threshold With and 1. The top section presents oral thresholds with nose clips on during
Without Nose Clips in Experiment 3A delivery followed by nose clip removal during assessment, enabling purely
retronasal olfaction, and the bottom section is the same as in Table 1.
Participant mM

Without nose clips the average headspace concentration of each sample. Although
1 0.0103
2 0.0056 measurements could not be made near the concentration of oral
3 0.0119 thresholds, we calculated the headspace concentration at oral
4 0.0039 threshold by extrapolating Henry’s law (P ⫽ x ⫻ K) to these
5 0.0075 concentrations (where P is vapor pressure, x is molar fraction, and
6 0.0087
K is a constant).
7 0.0077
8 0.0178 Although the liquid concentrations differed between nasal
9 0.0016 thresholds (3.42 ⫻10⫺3 M in PEG200 [Experiment 1]) and oral
10 0.0189 thresholds (9.00 ⫻ 10⫺6 M in water [Experiment 2]) by almost
Geometric M 0.0077 three orders of magnitude, the headspace concentrations were
With nose clips approximately the same (0.43 ␮g/L and 0.15 ␮g/L, respectively;
see Figure 5). The headspace concentration of the oral thresholds
1 0.2450 was slightly lower than the headspace concentration of nasal
2 0.0820 thresholds. There are three main possibilities that account for this
3 0.0291
4 0.0975
small difference. One possibility is that when samples were held in
5 0.1125 the mouth, the liquid temperature increased slightly because of the
6 0.0615 increased temperature of the mouth, relative to room temperature.
7 0.0581 If the liquid temperature increased, the headspace concentration
8 0.0388 would increase in proportion to liquid temperature. When 10 ml of
9 0.1263
10 0.0475 21° C water is introduced to the oral cavity for 30 s, the temper-
Geometric M 0.0746 ature increases to 32° C, as measured by a digital thermometer in
our laboratory. When aqueous menthol was heated to this temper-
Note. The geometric mean results of Experiment 3A for each participant ature, the headspace concentration was even closer to the nasal
comparing oral thresholds without nose clips, enabling both orthonasal and
retronasal olfaction (top section) and oral thresholds with nose clips,
threshold bottles’ headspace concentration. A second possibility is
disabling olfaction (bottom section). For comparison, the geometric mean that detection in the oral cavity was facilitated by the integration of
oral threshold from Experiment 2 is .009 mM. odor and taste. At suprathreshold levels, integration of the two
106 NAGATA, DALTON, DOOLITTLE, AND BRESLIN

Figure 5. The headspace concentrations of nasal (left panel) and oral (right panel) l-menthol thresholds against
the concentrations of l-menthol in liquid vehicle samples. In the left panel, the x-axis depicts the log molar
concentration of l-menthol in polyethylene glycol (PEG) 200, and in the right panel the x-axis depicts the log
millimolar concentration in deionized Millipore filtered water. In both panels, the y-axis represents the log
headspace concentration in ␮g/L. The vertical dashed line shows the geometric mean l-menthol thresholds
(3.420 mM, left panel; 0.009 mM, right panel). The horizontal dashed line shows the headspace concentration
of the threshold samples (0.43 ␮g/L, left panel; 0.15 ␮g/L, right panel).

modalities has been reported (Gillan, 1983; Murphy & Cain, Method
1979). Previous work in our laboratory has provided evidence in
Participants. Ten naive volunteers (5 women) were paid to participate.
humans of subthreshold, cross-modal chemosensory integration
Participants ranged in age from 19 to 36 years (average age ⫽ 23.7 years).
(Dalton, Doolittle, Nagata, & Breslin, 2000). When l-menthol Seven participants had previous experience in psychophysical experiments
samples are in the oral cavity, the l-menthol headspace concentra- but had not participated in these menthol studies.
tion may be slightly below the retronasal threshold concentration, Stimuli. Samples were the same as in Experiment 2.
but integration between the retronasal odor and the taste of men- Procedure. To determine the effect of retronasal olfaction on oral
thol could enable the increased absolute sensitivity. And third, the thresholds for l-menthol, we compared thresholds under two conditions. In
orthonasal and retronasal olfactory thresholds may simply differ one condition, retronasal and orthonasal flow was blocked; in the second,
slightly because of the exposure to different epithelial surfaces retronasal flow was permitted only while the stimulus was in the mouth. In
from the two directions of flow in the airway. Detection of the first condition, participants were given two samples (10 mL each) in
l-menthol using oral thresholds occurred at much lower concen- 40-mL plastic medicine cups. They were instructed to hold the nares closed
with foam rubber nose clips, rinse, take the first sample, hold it in their
trations when thresholds were measured without nose clips, thus
mouth for 5 s, expectorate, and rinse. They repeated the procedure with the
implicating retronasal sensory cues as an important determinant of second sample and judged which sample contained the stimulus. Cups were
the sensitivity to l-menthol when presented orally. presented in random order. A 45-s ITI was exercised, during which the
nose clips were removed and participants were allowed to breathe
Experiment 3B: Oral Detection Thresholds With the Nose normally.
Closed or Open for Retronasal Olfaction In the second condition, participants were given two samples (10 mL
each) in 40-mL plastic medicine cups. They were instructed to hold the
In Experiment 3A, the nose-open condition allowed retronasal nares closed with foam rubber nose clips, rinse, take the first sample into
olfaction to occur, which is a major component of oral-based the mouth, remove the nose clips, and hold it in their mouth for 5 s while
olfaction and food flavor. We assumed that the majority of olfac- breathing through the nose. After holding the sample in the mouth for 5 s,
tory input from oral stimuli was using this pharyngeal pathway. participants were instructed to replace the nose clips, expectorate, and
rinse. They repeated the procedure with the second sample and judged
We did not, however, explicitly eliminate orthonasal input. As the
which sample contained the stimulus. L-menthol and blank cups were
menthol solution was drawn into the oral cavity, it must first pass
presented in random order. A 45-s ITI was exercised, during which the
under the open nose. This creates an opportunity for brief but nose clips were removed. Participants were also instructed to breathe
possibility sufficient sampling of volatiles for detection as the oral normally during this period. On each of 3 days, participants were tested
sample is taken. Our observations indicate that, under most in- once in each condition with a 15-min break between conditions. Conditions
stances, slight nasal inspiration of air accompanies the solution were counterbalanced among participants and visits.
being drawn into the oral cavity. Thus, volatiles may be sampled
even as the liquid bolus passes the lips. Therefore, we conducted Results and Discussion
Experiment 3B expressly to mirror Experiment 3A but to eliminate
orthonasal input and to compare detection of menthol without any Thresholds for oral l-menthol obtained with the nares closed
olfaction to detection with only retronasal olfaction. (1.35 ⫻10⫺4 M) were significantly greater ( p ⬍ .01) than thresh-
PSYCHOPHYSICAL ISOLATION OF SENSORY MODALITY 107

olds obtained with retronasal flow permitted (6.36 ⫻10⫺5 M; see Stimuli. An eight-bottle series ranging in concentration from 7.90 ⫻
Table 2). Thus, these results confirm the general conclusion of 10⫺4 M to 0.203 M in one quarter log steps was prepared in the same
Experiment 3A, that the detection of l-menthol in the oral cavity is manner as described in Experiment 1 by dissolving l-menthol in PEG200.
based on volatiles entering the nose rather than orally derived Procedure. On each trial, participants were presented with only one
sensory input. Although Experiment 3A permitted both orthonasal pair of bottles. One bottle contained l-menthol in PEG200; the other bottle
contained only PEG200. The participants simultaneously sniffed the pair
and retronasal olfaction in the nose-open condition, the present
and then decided which bottle (left or right) contained the stimulus.
experiment permitted only retronasal olfaction during oral sam-
Thresholds were collected using the method of fixed stimuli with an
pling. As in Experiment 3A, the oral thresholds in Experiment 3B ascending staircase beginning with the lowest concentration (7.90 ⫻ 10⫺4
both without olfaction and with retronasal olfaction showed little M). Testing continued until five correct responses were reported at a given
relationship (r2 ⫽ .38; positive trend). This weak trend is in the concentration. This concentration was considered the lateralization thresh-
opposite direction of the trend seen in Experiment 3A. Taken old concentration. Two ITIs (2 and 5 min) were again manipulated to test
together these two experiments strongly suggest that both of these for effects of adaptation and sensitization of the chemesthetic properties of
types of oral l-menthol thresholds (nose open and closed) are based l-menthol. Each participant was tested twice with both ITIs.
on independent mechanisms.
The difference in sensitivity between the nose-closed and the
nose-open conditions in Experiment 3A was almost tenfold, Results and Discussion
whereas the difference between the two conditions in Experiment
In Experiment 3B, we observed retronasal olfaction to be more
3B was slightly more than twofold. Therefore, it appears that
sensitive than orally restricted sensitivity to l-menthol samples.
olfactory detection is more sensitive when utilizing both orthona-
The question remained, however, whether the retronasal detection
sal and retronasal pathways than when only retronasal pathways
of l-menthol was based on olfactory or chemesthetic sensations. In
are engaged. Because different participants were used in Experi-
this experiment, we measured chemesthetic thresholds in the nasal
ments 3A and 3B and they differed slightly in their sensitivities to
cavity using the nasal lateralization technique (Wysocki et al.,
l-menthol with the nares closed (Experiment 3B participants were
1987, 1992) to compare them with the nasal detection thresholds
1.75 times less sensitive), it is difficult to draw absolute conclu-
sions regarding the relative roles of orthonasal olfaction and ret- obtained in Experiment 1.
ronasal olfaction when sampling a solution orally. It is possible The results are given in Table 3 for those who also participated
that orthonasal and retronasal inputs combine additively to in- in Experiment 1. For all participants, the lateralization thresholds
crease olfactory signal strength, but the present studies do not (2-min and 5-min ITIs) were one to two orders of magnitude
permit this conclusion. Regardless, because Experiment 3B par- higher than the nasal detection thresholds. The ITI was also varied
ticipants were 4.5 times less sensitive than the Experiment 3A in the lateralization procedure in order to examine the effect of
participants in the nose-open condition relative to differences in repeated presentations of high concentrations of l-menthol on
their nose-closed sensitivities, we can conclude that orthonasal chemesthetic sensitivity. During one test session, thresholds were
sampling of volatiles when drawing a solution into the oral cavity obtained with a 2-min ITI and during another session, they were
also plays a role in detecting oral l-menthol solutions. obtained with a 5-min ITI. Overall, the results showed no differ-
ence in sensitivity based on the different ITIs (8.00 ⫻ 10⫺2 M for
a 5-min ITI vs. 9.00 ⫻ 10⫺2 M for a 2-min ITI).
Experiment 4: Nasal Lateralization Threshold If we accept the premise that lateralization thresholds depend on
the participant’s ability to detect chemesthetic sensations in the
In the nasal cavity, l-menthol can elicit odor and chemesthetic
nasal passages and that pure odor signals cannot be localized in the
sensations; however, it is not clear by which modality absolute
nose (Kobal et al., 1989), then it appears that nasal detection of
detection in the nasal cavity is based. Experiments 3A and 3B
l-menthol (both orthonasal and retronasal) was based on odor
demonstrate that l-menthol detection is based on volatile menthol
sensations and occurred at far lower concentrations than those
entering the nose but do not identify whether nasal olfaction or
nasal somatosensation is involved in the detection. Nasal lateral- necessary to permit nasal lateralization.
ization has been a useful technique for measuring chemesthetic
thresholds in the presence of odor sensations (Cometto-Muñiz &
Cain, 1995; Wysocki, Dalton, Brody, & Lawley, 1987; Wysocki, Table 3
Green, & Malia, 1992). Evidently, only somatosensory signals can Geometric Means of L-Menthol in Polyethylene Glycol 200
be lateralized within the two nares. Thus, if the lateralization
threshold is much higher than the absolute nasal detection thresh- Nasal lateralization
old, then nasal detection can be deduced to be olfactory. If later- threshold
Nasal detection
alization thresholds in the nose are similar to absolute detection,
Participant threshold 5-min ITI 2-min ITI
then nasal detection can be deduced to be somatosensory.
1 0.0016 0.0762 0.1365
2 0.0008 0.1524 0.0683
Method 3 0.0010 0.1016 0.0254
4 0.0016 0.0190 0.1360
Participants. Six adult volunteers (4 women and 2 men) were paid to
participate. Participants ranged in age from 18 to 32 years (average age ⫽ Note. This table compares geometric mean results of Experiment 4 for
25.3 years). All participants had previous experience in psychophysical nasal absolute detection thresholds and nasal lateralization thresholds for
experiments, and 4 participants had participated in Experiments 2 and 3. each participant. ITI ⫽ intertrial interval.
108 NAGATA, DALTON, DOOLITTLE, AND BRESLIN

Experiment 5: Nasal Detection and Lateralization order effects are responsible for the difference seen here between
Thresholds Measured Concurrently threshold types because responses were highly stable in Experi-
ment 1 (see Figure 2) when only a brief pause was inserted
It is difficult to directly compare the values obtained in olfactory between two threshold tests.
detection tasks with those obtained in lateralization tasks because Although the threshold protocol used in Experiment 4 is much
of the increased cognitive demands imposed by the task of odorant simpler than that used in the other experiments, it nevertheless
lateralization. For example, in both tasks the individual is (explic- resulted in lateralization values very similar to those obtained in
itly or implicitly) asked to detect whether the stimulus is present. Experiment 5 using a modified staircase with two pairs of bottles
However, in the lateralization task, the individual is also required to identify both olfactory and chemesthetic thresholds concur-
to indicate the location of the odorant (right or left nostril). Thus, rently. Therefore, Experiment 5 effectively validates the simpler
the higher thresholds that are typically obtained for odorant later- and more commonly used method of nasal lateralization by as-
alization could potentially result from the additional cognitive cending staircase of fixed stimuli (Experiment 4).
demand imposed by this task. However, in a recent study that
obtained odor thresholds and lateralization thresholds in exactly
the same manner (i.e., asking participants to detect and localize the General Discussion
odor–irritant on each trial), participants were able to detect odor at The objective of this study was to illustrate how to psychophys-
much lower concentrations than they were able to accurately ically isolate the modality by which a multimodal stimulus is
lateralize it (Radil & Wysocki, 2000; Wysocki et al., 1992). detected. Given that the detection and perception of most stimuli
The method used in Experiment 5 was designed to minimize the involves multiple sensory modalities and/or submodalities, this
cognitive demand differences between the detection and lateral- approach has broad application, especially for gustatory, olfactory,
ization thresholds. The participant was asked the same question and somatosensory chemical stimuli, as well as more traditional
throughout the entire study (Which bottle contains the stimulus?) somatosensory and visual stimuli. For this purpose, we selected a
using a single threshold protocol regardless of the type of threshold model stimulus, l-menthol, which can act on three sensory modal-
being collected. ities but for which the basis of absolute detection was previously
unknown. The experiments reported here demonstrate that humans
Method detect l-menthol in the upper airways using olfaction and not by
Participants. Eight adult volunteers (3 women and 5 men) ranging in
taste or somatosensations, regardless of whether it is delivered
age from 20 to 33 years (average age ⫽ 25.6 years) were paid to nasally or orally.
participate. Frequency histograms of absolute detection thresholds for
Stimuli. A nasal series of l-menthol in PEG200 was prepared using the l-menthol in the nose and the mouth were unimodally distributed.
same method described for Experiment 1. The distributions of nasal and oral detection thresholds had high
Procedure. Just as in the detection threshold measurement from Ex- test–retest reliability. The observation that menthol detection
periment 1, participants were presented with two pairs of glass bottles on thresholds are unimodally distributed in the sample population
each trial. One pair contained l-menthol and a blank; the other pair suggests that sensitivity for menthol has multiple determinants. A
contained two blanks only. The participants simultaneously sniffed from unimodal distribution is traditionally interpreted to mean that
both bottles within a pair and sniffed from both pairs sequentially, then
multiple factors with additive effects create a continuum of vari-
decided which individual bottle contained the odorant. Participants fol-
lowed this method throughout the entire experiment. As in Experiments
ability within the population. Conversely, a bimodal or trimodal
1–3, a modified staircase procedure with a four-down, one-up, five-reversal distribution is often interpreted to mean that one or two variations
criteria was used. in a single factor, such as a l-menthol receptor, is separating the
First, an absolute detection threshold was measured, followed by a population into two or three clusters as a function of the number of
lateralization threshold. To collect the absolute detection thresholds, we variants. The variability of nasal (orthonasal) and oral (retronasal)
made movement up or down in concentration depending on whether the thresholds across participants spanned two to three orders of
participants correctly identified the pair of bottles containing the l-menthol magnitude. Thus, some participants may be 1,000 times more
when they reported which individual bottle they believed contained the sensitive than others at detecting l-menthol. The variability was
stimulus. After five reversals were achieved for the detection threshold, the somewhat smaller when menthol was detected retronasally with
lateralization testing would begin at the concentration of the detection
oral delivery, but this is likely due to low sampling. We measured
threshold. Movement, up or down in concentration, was then made de-
pending on whether the participant correctly identified the individual bottle
200 nasal thresholds from 100 participants and only 59 oral–
containing the l-menthol. retronasal thresholds from 20 participants in Experiments 3A and
3B. Variability for oral thresholds may well have increased if 10
times more participants were tested.
Results and Discussion
When retronasal flow was blocked by nose clips (Experiments
The results are very similar to the findings reported for Exper- 3A and 3B), thresholds for oral l-menthol were significantly higher
iments 1 and 4, for odor and lateralization thresholds, respectively. in concentration than oral thresholds without nose clips, indicating
The mean absolute detection threshold concentration in Experi- that olfaction can be an important sensory cue during testing of
ment 5 was 2.00 ⫻10⫺3 M (very similar to the Experiment 1 sensitivity for orally presented l-menthol. This was true whether
results), and the mean lateralization threshold concentration was olfactory input from the oral cavity was orthonasal and retronasal
9.00 ⫻ 10⫺2 M (virtually identical to the Experiment 4 results). or merely retronasal. When olfaction was restricted to the retro-
For every participant tested, the lateralization threshold was higher nasal pathway, however, sensitivity was lower than when both
than the detection threshold concentration. It is very unlikely that olfactory pathways were permitted, which suggests that volatiles
PSYCHOPHYSICAL ISOLATION OF SENSORY MODALITY 109

are sampled orthonasally from solutions as they are sipped and this References
pathway increases olfactory sensitivity. The chemesthetic lateral-
ization thresholds from two methods (Experiments 4 and 5) were Breslin, P. A. S., Gilmore, M. M., Beauchamp, G. K., & Green, B. G.
higher in concentration than the nasal absolute detection thresh- (1993). Psychophysical evidence that oral astringency is a tactile sensa-
olds, suggesting that l-menthol can be detected in the nose at tion. Chemical Senses, 18, 405– 417.
Cliff, M. A., & Green, B. G. (1993). Sensory irritation and coolness
concentrations that do not permit chemesthesis. Gas chromato-
produced by menthol: Evidence for selective desensitization of irritation.
graphic analysis indicates that both oral and orthonasal l-menthol
Physiology & Behavior, 56, 1021–1029.
thresholds were very similar with respect to headspace concentra- Cliff, M. A., & Green, B. G. (1996). Sensitization and desensitization to
tion of l-menthol, differing by less than a factor of three (0.43 and capsaicin and menthol in the oral cavity: Interactions and individual
0.15 ␮g/L), whereas their solution concentrations differed by differences. Physiology & Behavior, 59, 487– 494.
almost three orders of magnitude. These results strengthen our Cometto-Muñiz, J. E., & Cain, W. S. (1995). Relative sensitivity of the
conclusion that the absolute detection of l-menthol, both nasally ocular trigeminal, nasal trigeminal and olfactory systems to airborne
and orally, was based on a volatile compound in the nose, rather chemicals. Chemical Senses, 20, 191–198.
than l-menthol’s more poignant somatosensory qualities of sting- Cruz, A., & Green B. G. (2000, February 24). Thermal stimulation of taste.
ing and cooling in the mouth. The failure of lateralization threshold Nature, 403, 889 – 892.
concentrations to match absolute detection threshold concentra- Dalton, P. N., Doolittle, N., Nagata, H., & Breslin, P. A. S. (2000). The
tions suggests that somatosensory stinging and cooling in the nose merging of the senses: Sub-threshold integration of taste and smell.
Nature Neuroscience, 3, 431– 432.
was not the basis of detection either. Detection of l-menthol was
Delwiche, J. F., Lera, M. F., & Breslin, P. A. S. (2000). Selective removal
olfactory, regardless of the route of delivery.
of a target stimulus localized by taste in humans. Chemical Senses, 25,
Given that many stimuli act on more than one sensory modality, 181–187.
it is often not obvious which sense subserves absolute detection. Finger, T. E., Bottger, B., Hansen, A., Anderson, K. T., Alimohammadi,
For chemical stimuli, oral and nasal sensations can be isolated with H., & Silver, W. L. (2003). Solitary chemoreceptor cells in the nasal
the use of nose clips to block all ortho- and retronasal flow of cavity serve as sentinels of respiration. Proceedings of the National
volatiles. In the case of olfaction and nasal irritation, lateralization Academy of Sciences, USA, 100, 8981– 8986.
is a powerful technique because pure olfactory stimuli are not Gillan, D. J. (1983). Taste–taste, odor– odor, and taste– odor mixtures:
lateralizable (Kobal et al., 1989; Radil & Wysocki, 2000; Wysocki Greater suppression within than between modalities. Perception & Psy-
et al., 1992). Oral–lingual stimuli, however, cannot be separated chophysics, 33, 183–185.
with the same techniques, because gustatory stimuli are localizable Green, B. G. (1985). Menthol modulates oral sensations of warmth and
cold. Physiology & Behavior, 35, 427– 434.
as a function of gustatory qualities without discriminative somato-
Green, B. G. (1986). Menthol inhibits the perception of warmth. Physiol-
sensory cues (Delwiche, Lera, & Breslin, 2000; Shikata, Mc-
ogy & Behavior, 38, 833– 838.
Mahon, & Breslin, 2000). In addition, it is difficult to deliver an Green, B. G. (1992). The sensory effects of l-menthol on human skin.
oral chemical stimulus without concomitant tactile stimulation in Somatosensory and Motor Research, 9, 235–244.
the same areas. Nevertheless, l-menthol, as shown by this study, Kobal, G., Van Toller, S., & Hummel, T. (1989). Is there directional
serves as an excellent example of how the contribution of multiple smelling? Experimentia, 45, 130 –131.
sensory modalities, stimulated by a single compound, can be Liu, L., & Simon S. A. (1998). Responses of cultured rat trigeminal
analytically studied and isolated using a combination of psycho- ganglion neurons to bitter tastants. Chemical Senses, 23, 125–130.
physical techniques. Murphy, C., & Cain W. S. (1979). Taste and olfaction: Independence vs.
Although there are clear evolutionary benefits for sensory sys- interaction. Physiology & Behavior, 24, 601– 605.
tems to operate in combination to perceive external stimuli through Radil, T., & Wysocki, C. J. (2000). Butanol detection and lateralization:
multiple modalities and submodalities, the ability to decompose Conscious and unconscious mechanisms [Abstract]. Chemical Senses,
25, 624.
the gestalt percept in order to identify the basis for stimulus
Shikata, H., McMahon, D. B. T., & Breslin, P. A. S. (2000). Psychophysics
detectability is an important tool for (a) evaluating the degree of of taste lateralization on anterior tongue. Perception & Psychophysics,
additivity or synergy among sensory modalities and (b) examining 62, 684 – 694.
how the brain integrates inputs from different sensory systems Stevens, J. C. (1995). Detection of heteroquality taste mixture. Perception
(Dalton et al., 2000). This process of psychophysical isolation of & Psychophysics, 57, 18 –26.
sensory subsystems is also applicable to the perception of complex Stevens, J. C. (1997). Detection of very complex taste mixtures: Generous
stimuli that are not chemical in origin. Other examples of stimulus integration across constituent compounds. Physiology & Behavior, 62,
detection in which the specific roles of subsystems are unclear 1137–1143.
include visual identification– detection of an object, which could Wysocki, C. J., Dalton, P., Brody, M. J., & Lawley, H. J. (1987). Acetone
involve shape, color, shading, stereoscopic cues, and so forth, and odor and irritation thresholds obtained from acetone-exposed factory
the detection of cold water, which could include systems ranging workers and from control (occupationally non-exposed) subjects. Amer-
ican Industrial Hygiene Association Journal, 58, 704 –712.
from touch, to cool, to polymodal nociceptors. Which subsystems
Wysocki, C. J., Green, B. G., & Malia, T. P. (1992). Monorhinal stimu-
are used in the initial detection of the object–stimulus could be
lation as a method for differentiating between thresholds for irritation
studied using similar analytical approaches to the one demon- and odor [Abstract]. Chemical Senses, 17, 722–723.
strated in the present study with l-menthol. The strategies used in
the present study to psychophysically isolate and determine the
sensitivity of each modality contributing to the perception of a Received February 20, 2002
multimodal stimulus can have broad applicability in multisensory Revision received April 9, 2004
research. Accepted July 6, 2004 䡲

You might also like