Physiology& Behavior. Vol. 48, pp. 87-90. ©PergamonPress plc, 1990. Printedin the U.S.A. 0031-9384/90$3.00 + .

00

Responsiveness to Olfactory Stimuli

Presented in Sleep 1

PIETRO BADIA, 2 NANCY WESENSTEN, WILLIAM LAMMERS,

JOEL C U L P E P P E R A N D J O H N H A R S H *

Bowling Green State University and *University of Southern Mississippi

Received 18 September 1989

BADIA, P., N. WESENSTEN, W. LAMMERS, J. CULPEPPER AND J. HARSH. Responsiveness to olfactory stimulipresented in
sleep. PHYSIOL BEHAV 48(1) 87-90, 1990.--Whether humans react to olfactory stimuli presented in sleep was assessed. Responses
of ten participants (mean age = 22.8 years) were recorded to repeated three-minute periods of either air alone or to a peppermint odor
(0.26 mg/liter) during stage 2 sleep. These responses included behavioral (awakening, microswitch closure), autonomic (heart rate,
EMG, respiration), and central (EEG) components. An odor delivery system is described comprised of an aquarium pump, Teflon and
TYGON tubing, oxygen mask, filtering, and air flow valves. The data indicate that humans react behaviorally, autonomically and
centrally to olfactory stimuli presented while sleeping. Although the percentage of overall responsivity to olfactory stimuli was low,
significant differences (ANOVA) in responsivity to odor periods vs. nonodor periods were found for microswitch closures, EEG,
EMG, and heart rate. For these measures eight or more of the ten participants showed this pattern of differential responsivity during
odor and nonodor periods (Sign test =p<0.05). A time-of-night effect was also observed in that responsivity tended to be greatest early
in the night. The effect on responsivity of other durations, concentrations, and odors requires additional research.

Olfaction in sleep Responsiveness in sleep Odor and heart rate Odor effects on EEG Humans

THE present study assesses olfaction in sleep and reports behav-
ioral and psychophysiological responsiveness to a fragrance pre-
sented periodically during the night. It also describes a procedure
for presenting olfactory stimuli in sleep.
Recent reports by Schwartz and colleagues (7, 8, 11) indicate
that changes occur in cardiovascular, respiratory and electroen-
cephalographic (EEG) measures when odors are presented in the
waking state. These data suggest that psychophysiological indices
may be useful for determining whether the olfactory system is
functional in sleep. There are no known reports investigating
olfactory sensitivity in human sleep, but anecdotal accounts of
insensitivity to olfactory stimuli in sleep abound. In contrast to
olfactory stimuli, sensitivity of humans to auditory and visual
stimuli are well documented (2, 3, 5, 10). We should note that in
nonhumans there is some evidence in sleeping cats that olfactory
stimuli will desynchronize high voltage slow waves (5).
Finding the olfactory sense functional in sleep should encour-
age a more detailed analysis of olfaction and sleep in subsequent
studies. It may also provide information of practical value. For
example, awakening threshold of different odors could be deter-
mined which might have implications for safety-related consider-
ations (e.g., detection of smoke or gas). A functional olfactory
sense in sleep also would permit assessing the effects of different
fragrances on sleep quality. Odors are known to have both
subjective (alerting/relaxing, pleasant/unpleasant) and psycho-
physiological effects in the waking state (7, 8, 11). It may be that
odors reported as "relaxing" enhance sleep quality while those
reported as "alerting" degrade sleep quality.
The following study is a first step. The central question
addressed was whether the olfactory sense functions in sleep. [One
previous study related to olfaction and sleep was found in the
literature (9), however, its methodology was seriously inadequate.
The study was performed on 3-day old infants but sleep and wake
were not objectively determined via polygraphic recording. Also,
a startle response measure was used which occurs spontaneously
with high frequency in young infants.] We describe below a
procedure for assessing this question, and report behavioral and
psychophysiological data related to it.
METHOD
Subjects
Subjects were 10 undergraduate and graduate students (6
females, 4 males, mean age=22.8) enrolled at Bowling Green
State University. They were screened prior to participation for
medication use, for sleep, hearing, health and allergy problems,
and for responsiveness to olfactory stimuli in the waking state.
Apparatus
Grass 7P511 AC amplifiers (Model 78D polygraph) were

~This research was funded by the Fragrance Research Fund Ltd., New York, NY.
2Requests for reprints should be addressed to Pietro Badia, Psychology Department, Bowling Green State University, Bowling Green, OH 43403.

87
 88
used to record electroencephalographic (EEG), electrooculographic
(EOG) and eletromyographic (EMG) signals (the standard sleep
montage was used) and a 7P4 preamplifier for heart rate; a Grass
Model PRT8 pneumatic transducer with a 7P3 preamplifier was
used for recording respiration. Button press responses were
monitored using a custom-built microswitch consisting of a
cylindrical tube (button on top) encased in a soft leather glove.
White noise (50 dB SPL) was generated by a Grason Stadler 455B
Noise Generator. An Apple lie + microcomputer interfaced with a
Cyborg Model 91A A/D converter controlled stimulus presenta-
tion and recorded heart rate and respiration rate.
Fragrance delivery system. The fragrance delivery system
consisted of a Whisper 1000 aquarium pump, charcoal air filter,
glass bottles (35 ml), air flow control valves, TYGON (7 mm) and
Teflon (2 mm) tubing, and an oxygen mask (B & F Medical--
Model 64009 Pediatric Oxygen Mask) for fragrance delivery. The
mask was slightly modified such that circular holes (25 mm in
diameter) were placed on each side of the mask. The peppermint
fragrance [supplied by International Flavors and Fragrances (IFF),
Union Beach, NJ] was embedded in pellets (2 mm) composed of
low density polyethylene plastic with 20% peppermint fragrance
oil. Odor concentration at the mask was controlled by the air flow
and the number of pellets used (IFF provided evaporation rates of
pellets). For example (present study), with an airflow of 0.5
liters/min and 35 pellets in a jar, the concentration of fragrance at
the mask was 0.26 mg/liter (___10%). Air flow generated by the
aquarium pump passed through a charcoal filter, into a series of
valves, and out to the oxygen mask (total travel distance from
control room to bedroom = approximately 20 ft). To prevent
fragrance contamination or "lingering fragrance," Teflon tubing
was used to transmit the fragrance. The teflon tubing was inserted
into TYGON tubing to protect the teflon from crinkling in the
event of marked movements in sleep. The air flow valves were
arranged so that the first set of valves was directly connected by
TYGON tubing. A bottle containing peppermint pellets (35)
intervened between the second set of valves. Thus, opening the
first set of valves (second set closed concurrently) resulted in
presentation of unscented air. Opening the second set of valves
(first set closed concurrently) resulted in presentation of the
fragrance. To avoid trigemlnal cues to air flow and to assure
evacuation of previous fragrances, a constant flow of air (0.5
litter/min) was presented through the mask at all times whether or
not a fragrance was presented. The fragrance concentration of 0.26
mg/liter was chosen because the subjective intensity of the odor in
waking was judged maximal at this level.
Procedure
Subjects arrived at the Sleep and Psychophysiology Laboratory
at 2230 hr, at which time they were given a description of the
procedure and informed consent was obtained. An olfactory
sensitivity test then was administered that required subjects to
identify four different fragrances (all subjects correctly identified
the fragrances). Next, electrodes were secured for recording EOG
(outer canthi, above and below midline), EMG (supra- and
submental), EEG (C3, C4, O1, 02, left and right mastoids), and
heart rate (modified Lead II). The respiration belt was secured
comfortably around the rib cage at about the 4th and 5th rib. The
oxygen mask was secured with a cloth elastic band.
Subjects slept in an electrically shielded, sound-attenuated
bedroom. They were told that a fragrance would be presented
several times during sleep, and that their task was to respond to the
flagrance by pressing the microswitch and to awaken as soon as
they detected the fragrance. They were told that if they failed to
respond, they would be awakened by the technician calling their
BADIA ET AL.
name over the intercom and repeating the instructions. To enhance
responsivity (2, 3, 5), subjects were in fact awakened and
instructions repeated following the first two failures to respond,
and once again later in the night.
Subjects were presented with five fragrance trials prior to sleep
to ensure that they were responding appropriately. An additional
five fragrance trials were given the following morning upon
awakening to assure continued integrity of the fragrance delivery
system.
Trials were presented in stages 2 and REM only. Both
fragrance trials and nonfragrance (baseline, air only) "trials" were
presented randomly with a 50% probability. Air flow through the
mask remained at 0.5 liter/min throughout the night. Fragrance
trials consisted of a three-minute presentation of peppermint odor
(concentration at the mask was 0.26 mg/liter). A peppermint
fragrance was chosen based on information provided by IFF that
this fragrance results in subjective reports of being both hedoni-
cally pleasant and also alerting. Nonfragrance "trials" (baseline)
consisted of a selected three-minute period of air only and
provided a means of comparing random responding in sleep to
responding in conjunction with fragrance presentation. The inter-
val between all trials (fragrance and nonfragrance), from the end
of one trial to the beginning of the next trial, was four minutes.
Subjects were monitored continuously (from "lights out" to
"lights on") via infrared video camera.
Psychophysiological/Behavioral Response Scoring
Several measures were scored for each of the three-minute
fragrance and nonfragrance periods. These measures included
occurrence of an awakening (>15 sec alpha), a microswitch
closure without an awakening, "speeding" in the EEG tracing
(see below), change in muscle tone (EMG), heart rate (HR), and
respiration rate. "Speeding" in the EEG tracing was defined as a
high frequency EEG burst (similar to awake EEG) lasting less than
10 sec. The frequency of occurrence for awakenings, microswitch
closures, EEG speeding, and EMG were converted to percentages
for statistical analyses [percent responding = number of times a
response occurred for a given condition/number of "opportuni-
ties" (i.e., three-min periods or trials) for which a response could
have occurred for that condition]. Respiration and HR were
averaged on a per minute basis.
Time-of-night recording. To assess differential responsivity
across the night (Table 1) each individual record was divided into
three equal portions of the night (first, second, third) based upon
the first and last epoch on which a trial was presented.
RESULTS
Design and Data Analysis
Only stage 2 data were analyzed since too few opportunities for
fragrance presentations occurred in stage REM. However, the
trends observed in stage REM were identical to those of stage 2.
Each dependent measure was analyzed separately for the three-
minute periods using a two-way Condition (fragrance, nonfra-
grance) × Time-of-Night repeated measures design (Biomed
BMDP2V program--Greenhouse-Geisser correction for degrees
of freedom). In addition, to compare the number of subjects
displaying greater responsivity under the fragrance condition than
under the nonfragrance condition, each dependent measure was
analyzed using the Sign test.
Behavioral Responsivity
Microswitch closure. Our first question was whether subjects
RESPONDING TO OLFACTORY STIMULI IN SLEEP 89

TABLE 1 of-Night nor the Condition × Time-of-Night was significant.

RESPONSIVITYUNDERFRAGRANCEAND NONFRAGRANCE Again, the effect of fragrance on EEG speeding was widespread
CONDITIONSACROSSTIMEOF NIGHT among subjects in that a greater percentage of EEG speeding
during fragrance trials was noted in nine of the ten subjects (Sign
Condition test =p<0.05).
Time
of Night Fragrance Nonfragrance Autonomic Responsivity
Heart rate. Whether psychophysiological responses [i.e., heart
Microswitch 1 16.74 3.37 rate, breathing rate and EMG increases (see below)] would be
Closure* 2 16.19 0.00 sensitive indices of responsivity to fragrances was also assessed
3 4.37 0.44 (Table 1). Heart rate tended to be higher under the fragrance
Awakening* 1 7.41 4.95 condition for the first part of the night. Since the condition ×
2 7.44 3.50 Time-of-Night interaction was marginally significant, F(2,18) =
3 4.00 2.94 3.31, p = 0 . 0 7 , the interaction was analyzed further by one-way
comparison of fragrance versus nonfragrance conditions for each
EEG 1 22.13 0.32 time of night and it was revealed that the higher HR under
Speeding* 2 16.37 5.61 fragrance was significant (p = 0.05) for the first part of the night.
3 8.31 5.62 An analysis of the number of individual subjects displaying a
EMG* 1 36.25 44.21 higher heart rate for fragrance trials than for nonfragrance trials
2 28.91 40.87 was performed using the Sign test. This analysis revealed that
3 37.45 46.87 eight of the ten subjects had a higher mean rate on fragrance trials
(p=0.05).
Average 1 63.84 60.96 Respiration. Breathing rate did not differ between fragrance
HR (bpm)~" 2 59.08 58.00 and nonfragrance trials for any time of night. However, a
3 57.19 57.18 Time-of-Night effect was observed in that breathing rate was
*Percentage of stimuli eliciting a response. higher during the first portion of the night (collapsed across
tAverage HR under fragrance and nonfragrance conditions. Condition) than during the second or third portions of the night,
F(2,18) = 6.72, p<0.05. No other effects were significant.
EMG increase. We also addressed whether the frequency of
phasic muscle tone (chin) changes was affected by the presentation
could make microswitch closures to olfactory stimuli presented
of fragrances (Table 1). Fragrance presentations had an inhibitory
during sleep. The data indicate that they can although the overall
effect on EMG activity in that the percentage of trials in which an
prercentage of responding was low. The percentage of micro-
EMG change occurred was 34.2 for fragrance trials compared to
switch closures while asleep was greater for fragrance trials than
43.9 for nonfragrance trials. A two-way Condition × Time-
for nonfragrance trials (Table 1). The latter was true expecially for
of-Night ANOVA revealed a significant main effect for Condition,
the In'st (16.74% vs. 3.4%) and second (16.19% vs. 0%) parts of F(1,9)=6.56, p<0.05, confmning the above observation. This
the night. When collapsed across all three times of night the means
pattern of responding was found in eight of the ten subjects and
for fragrance and nonfragrance periods were 12.43% and 1.27%, was statistically significant (Sign Test, p=0.05). Neither the
respectively. A two-way Condition × Time-of-Night ANOVA Time-of-Night main effect nor the interaction was significant
revealed a significant main effect for Condition, F(1,9)= 15.18,
(p>0.05).
p<0.05, confLrming this observation. The pattern of greater
responsivity on fragrance trials was generalized in that eight of the
ten subjects performed in this manner (Sign test=p=0.05). DISCUSSION
Neither the main effect for Time-of-Night nor its interaction with
Condition was significant (p>0.05). The following can be said regarding olfactory sensitivity in
Awakenings. Whether subjects could awaken to fragrances sleep. There were significantly more microswitch, heart rate and
presented during sleep was also of interest. Although the percent- EEG speeding responses to fragrances than to nonfragrances. In
age of awakenings was higher for fragrance trials than for contrast, significantly less EMG activity was observed during
nonfragrance trials for all times of night (Table 1), the two-way fragrance periods. No differences were found between fragrance
ANOVA for Condition × Time-of-Night failed to reveal any and nonfragrance periods for awakenings and respiration. Eight or
significant effects (p>0.05). Only six of the ten subjects had a more of the ten subjects displayed the pattern of responding
greater percentage of awakenings under fragrance trials. described for these response measures. A Time-of-Night effect
was also observed with some measures in that responsivity was
greater during the earlier portions of the night.
Electroencephalographic Responsivity
The data clearly indicate that humans can detect olfactory
EEG speeding. Whether electroencephalographic measures stimuli in sleep. We assume that detection of such stimuli is due to
would indicate reactions to a fragrance in sleep in the absence of stimulation of the olfactory sense. However, the present study
full awakenings was also assessed (Table 1). As can be seen, the cannot rule out the possibility of trigeminal chemoreception. Tests
percentage of EEG speeding (10 sec or less) was greater for of additional odors at different concentrations would provide
fragrance trials than for nonfragrance trials for all three portions of greater confidence that the olfactory modality was the primary
the night. A two-way ANOVA revealed a significant main effect route of detection. Be that as it may, the present data indicate that
for Condition, F(1,9)=5.88, p<0.05, confLrming the above sensitivity to olfactory stimuli presented in sleep can be assessed
observation that the percentage of EEG speeding was greater for using central (EEG), autonomic (HR, EMG), and behavioral
fragrance trials (15.60%) versus nonfragrance trials (3.85%), (microswitch) measures. However, reactivity of the different
collapsed across time of night. Neither the main effect of Time- measures varied markedly. The measures least likely to detect
90
olfactory sensitivity were awakenings and respiration changes.
The latter respiration findings are counterintuitive given that
respiration is affected in waking (11). Also surprising are the EMG
findings. While most response measures showed an increase in
activity, the EMG decreased. The finding of lower EMG activity
during fragrance periods compared to baseline periods is perplex-
ing. Clearly, additional data are needed before more can be said
concerning the finding. It is apparent that closure of the hand
microswitch did not affect the chin EMG. Had it done so EMG
activity would have been higher.
Compared to the auditory modality, responsivity to olfactory
stimulation is relatively low (2, 3, 5). However, it is inappropriate
to make such judgments regarding responsivity without benefit of
cross modal psychophysical scaling of the relevant sensory dimen-
sions. It is possible that other concentrations or other odors
(different hedonics) would result in higher levels of responsivity.
Also, the use of different response measures and procedures in
auditory studies may play a significant role in accounting for the
disparity in findings. We should note, however, that responses to
the olfactory stimuli occurred while sleeping, frequently without
subsequent waking, and many occurred in the absence of brief
bursts of EEG speeding.
One factor known to affect responsivity in sleep is the
contingency between appropriate responding and inappropriate
responding (i.e., failure to respond). Those auditory studies
acquiring marked control over responding systematically rein-
forced responding and punished failure to respond; those studies
REFERENCES
1. Badia, P.; Harsh, J.; Balkin, T.; Cantrell, P.; Klempert, A.; O'Rourke,
D.; Schoen, L. Behavioral control of respiratiori in sleep. Psycho-
physiology 21:494-500; 1984.
2. Badia, P.; Harsh, J.; Balkin, T.; O'Rourke, D.; Burton, S. Behavioral
control of respiration in sleep and sleepiness to signal induced sleep
fragmentation. Psychophysiology 22:517-524; 1985.
3. Badia, P.; Lammers, W. Responsivity in sleep and time-of-night.
Psychophysiology 25:433; 1988 (abstract).
4. Harsh, J.; Badia, P.; O'Rourke, D.; Burton, S.; Revis, C.; Magee, J.
Factors related to behavioral control by stimuli presented in sleep.
Psychophysiology 24:535-541; 1987.
5. Hemandez-Peon, R.; Lavin, A.; Alcocer-Cuaron, C.; Marcelin, J. P.
Electrical activity of the olfactory bulb during wakefulness and sleep.
EEG Clin. Neurophysioh 12:41-58; 1960.
6. Lammers, W.; Carpenter, G.; Harsh, J.; Badia, P. Responsivity in
BADIA E T AL.
acquiring less control did not do so (2, 3, 5). Although the present
study awakened subjects for not responding, awakenings occurred
only 3 times across the night. Presumably the use of a more
systematic and demanding procedure would have enhanced re-
sponsivity.
The time-of-night effect (i.e., less responsiveness in the latter
third of the night) is similar to that reported by those presenting
auditory stimuli in sleep [e.g., (4,6)]. This effect most likely
relates to the temperature trough which occurs during the latter
third of the night. A clear positive relationship has been reported
between responsiveness in sleep and core body temperature.
Similar effects have been reported in the waking state under sleep
deprivation.
Finding that the olfactory sense is functional in sleep permits
the study of sensitivity to other odors, including parametric
variation of odor concentrations and durations. It also permits the
study of the effects of odors on sleep quality. As noted earlier,
odors are known to have both subjective effects (alerting/relaxing,
pleasant/unpleasant) and psychophysiological effects in the wak-
ing state (EEG changes). It may be that odors reported as
"relaxing" will enhance sleep quality while those reported as
"alerting" will degrade sleep quality.
ACKNOWLEDGEMENTS
We want to thank Craig Warren and Frank Schiet of the International
Flavors and Fragrances R & D for their efforts in the development of the
fragrance delivery system.
sleep, temperature, and time-of-night. Sleep Res. 18:427; 1989
(abstract).
7. Lorig, T. S.; Schwartz, G. E. Brain and odor h Alteration of human
EEG by odor administration. Psychophysiology 24:599; 1987.
8. Lorig, T. S.; Schwartz, G. E. Brain and odor II: EEG alpha activity
during administration of perceptually similar odors. Psychophysiology
24:599; 1987.
9. Murray, B.; Campbell, D. Difference between olfactory thresholds in
two sleep states in the newborn infant. Psychonom. Sci. 18:313-314;
1970.
10. Okuma, T.; Nakamura, K.; Hayashi, A.; Fujimori, M. Psychophys-
iological study on the depth of sleep in normal human subjects.
Electroencephalogr. Clin. Neurophysiol. 21:140-147; 1966.
11~ Schwartz, G. E.; Whitehom, D.; Hernon, J. C.; Jones, M. Subjective
and respiratory differences of fragrances: Interactions with hedonics.
Psychophysiology 23:460; 1986.