Effects of Ambient Odors
on Slot-Machine Usage in
a Las Vegas Casino
Alan R. Hirsch
Smell & Taste Treatment and Research Foundation, Ltd.
ABSTRACT
Studies have shown that ambient aromas impact on consumers'
behavior. To further investigate such effects, for one weekend, two
slot-machine areas in a Las Vegas casino were odorized, eacll with a
different aroma. A third slot-machine area &el"W'ed as an unodorized
control. The amounts of money gambled in slot machines located in
the three areas were measured and eompared for the weekend of the
odorization and for the weekends before and after as well, to eontrol
for extraneous variables. Our data show that the amount of money
gambled in the slot machines surrounding Odorant No. 1 during the
experimental weekend WBB greater than the amount gambled in the
same area during the weekends before and after the experiment by
an average of45.ll% (p= <0.0001). Farther, the IIJilOUIItofthe
increase appeared greater on Saturday when the concentration of
odorant was higher: mean increase on Saturday was 153.42%
(p = <0.0001) versus JD.-n increase on Sunday of 33..66% (p = <O.OOS).
The amounts of money gambled in the slot machines ll11lT'Otllllii
Odo:rant No. 2 and in the control area did not cllaDge significantly
compared to the previous weekend and the weekend following the
experiment. The likely mechanjsm of action of the effective odo:rant
and the implications oftheae results are discussed. ®1995 John
W'dey & Sons, Inc.
From conditioned-re:flex experiments with animals not known to possess conscious processes, it seems clear that consciousness is not
Psychology & Marketing
© 1995 John Wuey & Sons, Inc.
Vol. 12(7):585-594 (October 1995)
CCC 0742-60461!151070585-10
585
l'd
60L9·908·l88
J86UISU98 J960tj
dgg:ll Ol OG uer
required for a conditioned response. Behavioral responses and adaptations triggered by stimuli to the neural biological systems through unconscious processes are preserved by evolution. But in the case of
humans, with our anatomically advanced neurological mechanisms,
such unconscious processes may become conscious as emotions or subjective feeling states.
With increasingly sophisticated understanding of the impact of
emotions and subjective feelings on consumers' behavior, studied efforts are made in retail marketing to enhance consumers' transient
affective states through appeala to the senses of sight, sound, and
touch. Appeals to the sense of smell have been utilized only to a minor
extent, perhaps ironically, because of the human senses, the sense of
smell has the greatest impact on emotion. Attesting anatomically to
this, the nose is directly connected with the olfactory lobe which is
part and parcel of the limbic system, that area ofthe brain considered
the seat of the emotions (MacLean, 1973).
Profuse neuronal pathways connect the nose and olfactory structures to the prepiriform cortex, the emotional center of the brain, to
the amygdala, long considered important in various kinds of emotional behavior, to the hypocampus, another important functional
component of the limbic system, and to numerous other limbic structures.
These olfactory-limbic interconnections help to explain the phenomenon called olfactory-evoked recall, which occurs when an odor
stimulates the vivid recall of a memory or scene from the past.
In a 1989 study, Knasko showed that ambient aroma had an impact
on the duration of time consumers spent at a jewelry counter. Our
1991 study (Hirsch & Gay) showed that consumers were more likely
to purchase a well-known brand of athletic shoes displayed in a perfumed room than identical shoes displayed in an unodorized, but otherwise identical room. In the shoe study, results were based on
consumers' subjective responses concerning preferences and intentions rather than on actual purchases, so we sought further to investigate the effects of ambient aroma on spending behavior in an area
that could be measured objectively.
One area where amounts of money spent could easily be tracked
and measured, and where no inadvertent effects due to salespersons'
knowledge of an ongoing experiment would be a factor, is gambling in
slot machines. Further, the amount of money spent gambling appears
to be less affected by external variables such as sales, Christmas, and
other seasonal promotions that influence spending behavior in stores.
In casinos it is common to incite patrons to gamble by directly appealing to their senses: the exciting sounds of coins jingling, sirens
screaming when someone hits the jackpot, intense lighting, plush carpets, luxurious surroundings, and controlled temperature. But these
enticements are uniformly present at all times and all seasons.
586
(;'d
HIRSCH
60L9·908- ~ 88
J86UISU98 J960tj
dgg:~~ o~
02: uer
Therefore, we chose gambling in slot machines as our paradigm for
this study, which was approved by the Institutional Review Board.
MEmODS
Two pleasant but distinctly different odorants were chosen based on
earlier preference surveys. The site of the experiment was the casino
ftoor of a large hotel in Las Vegas. Odorant No. 1 was placed in dispensing devices amid 18 quarter slot machines. Odorant No. 2 was
placed in dispensers in another gambling area amid 19 nickel and 9
quarter slot machines. A control area with 22 quarter slot machines
was not odorized. Figure 1 is a map of the casino floor showing placement of both odorants and the control area.
Odorization began at midnight Friday and proceeded for 48 hours
through Saturday and Sunday. Both odorants were ejected at
suprathreshold levels (above the threshold of awareness), but at somewhat lesser levels on the second day, Sunday.
The amount of money gambled was measured in dollars and cents
at each slot machine on the Saturday and the Sunday of the experimental weekend and on the Saturday and Sunday of the previous
weekend and the following weekend as well. This was done to assess
any spurious effects such as weather, presence of conventions, and
floor shows. Figures for an average weekend also were available for
comparison.
Sto.tistical Analysis. Data were analyzed using the student's t test
for the difference of independent means to assess the effects of the
odorants on the amounts gambled.
RESULTS
The amounts of money gambled in the area of Odorant No.1 increased
significantly during the weekend of the experiment. Table 1 shows the
percent increase at each individual slot machine compared to the previous weekend and also compared to an average weekend. Table 1 also
shows the percent decrease in amounts gambled in each slot machine
after Odorant No. 1 was removed.
Table 2 shows that the overall amount of money gambled in the area
of Odorant No. 1 increased during the experimental weekend (second
weekend) by 45.11% (p= <.0001) compared to the overall amounts for
the weekends before (first weekend) and after (third weekend).
Looking at the data for each day separately: Table 3 shows the percent changes in amounts gambled at each slot machine on Saturday
when Odorant No. 1 was added compared to the amounts gambled the
Al\IBIENT ODORS AND SLOT-MACHINE USAGE
s·d
60L9·908· ~ 88
o87
J86UISU98 J960tj
dgg:~~ o~ OG
uer
Jan 20 10 11 :58p
Roger Bensinger
HOSHIH
...
831-305-6709
p.4
Table 1. Percent Change in Amounts Gambled During the Weekend
Due to tbe Presence of Odorant No. 1.
Percent Increase When
Odorant Was Added
SlotMaclrine
Over Previous Weekend
1
Percent Increase Over
Average Weekend
4
5
6
7
8
84.52
138.03
67.99
32.65
10.38
85.59
47.53
170.08
75.60
137.23
34.60
160.43
79.95
76.77
114.68
210.80
13.04
20.36
9
10
59.87
63.33
57.88
27.46
62.53
33.76
-0.54
37.44
71.43
89.10
11
12
13
14
15
16
17
18
-134.20
-113.00
-105.64
-74.41
-28.21
-24.54
-28.73
-78.81
-194.80
-29.50
-137.02
-31.49
-154.29
-122.71
-693.63
-226.77
54.88
13.56
38.56
64.61
38.28
27.23
-0.89
69.89
2
3
Percent
Decrease After
Odorant No. 1
Was Removed
-179.45
41.31
Table 2. Overall Percent Clumge in Amount Gamhled-DataAnal.ymd by
Student's t That of Difference of Independent Means.
First versus seeood weekend
Mean difference
Significance
First versus third weekend
Mean dill:erenee
Significance
Seeood versus third weekend
Mean difference
Significance
Average of first and third weekends
versus second
Odor No.1
Control
Odor No.2
Area
Area
Area
83.37%
p<.0001
16.38%
17.84%
1.31%
<NSJ
-4.4%
-44.75%
p<.0001
-17.89%
<NSJ
2.33%
(NSJ
45.11%
p<.0001
15.98%
(NSl
6.41%
CNSJ
(NS)
(NS)
(NS)
20.59%
<NS)
Nqte Second. weekend was the experimental weebod. FiTst and third weekends were without intervention.
NS=not~t.
AMBIENT ODORS AND SLOT-MACHINE USAGE
g·d
60L9·908· ~ 88
589
J86UISU98 J960tj
d5g:~~ o~
OG uer
Table 3. Percent Change In Amounts Gambled on Saturday
Due to tbe Presence of Odonmt No. 1.
Slot Machine
Percent
Decrease the
Percent Increase When
Odorant Wu Added
Percent Increase Over Saturday Odorant
No. 1 WM Removed
Over Previoll8 Saturday
Average Saturday
18.03
50.40
68.94
54.83
19.52
35.20
79.28
36.44
1
2
3
4
5
6
7
8
9
10
8Q.42
73.51
70.88
47.63
58.83
11
12
13
14
15
16
17
18
47.99
56.90
24.18
62.00
73.58
-230.86
-155.45
-210.38
-112.93
6.74
-86.35
-13.53
-54.25
-249.01
-9.67
-242.12
-197.02
-98.45
-112.24
-1461.06
-234.96
-334.29
32.48
78.26
125.37
216.08
117.08
6.55
68.83
83.83
55.78
314.62
70.09
242.78
132.44
118.44
101.75
303.97
89.26
227.71
14.60
Table4. Peroont Change In Amount Gambled on Satnrday-DataAnalyzed
by Student's t 'Thst of Difference of Independent Means.
Firat versus seoond Satorday
M-ean ditlerence
Significance
First versus third Saturday
Mean difference
Significance
Second versus third Saturday
Mean difference
Significance
Odor No.1
Area
Control
Area
Ode>rNo. 2
Area
125.3%
p < .0001
40.3%
CNS)
56.6%
<NSJ
9.9%
(NS)
28.4%
CNS)
54.8%
(NSJ
-51.2%
-8.5%
(NSJ
(NS)
18.6%
(NSJ
18.65%
<NSl
p < .0001
1.2%
Average of first and third Saturdays
versus second
Mean d:i:ffurew:e
Significance
590
g·d
53.42%
p < .0001
HIRSCH
60L9·908· ~ 88
J86UISU98 J960tj
d5g:~~ o~ OG
uer
previous Saturday and the amounts on an average Saturday. It also
shows the decrease in the amounts gambled at each slot machine on
the Saturday after Odorant No. 1 was removed. The overall amount
gambled in the area of Odorant No. 1 increased on Saturday by
53.42% (p = < .0001). See Table 4.
Tables 5 and 6 show similar data for Sunday when the amount
gambled in the area of Odorant No. 1 increased by 33.6%
(p= <.003).
Further, a comparison of the amounts gambled during the week
prior to the study and the week after the study (first vs. third weekends) shows that they were not significantly different. This suggests
that the baseline data had low variability and that the increase in
amounts gambled during the experimental weekend (second weekend) was, in fact, due to the presence of Odorant No. 1 (Tables 2, 4,
and 6).
The amounts gambled in the area with Odorant No. 2 and in the
control area during the experimental weekend were not significantly
different from the amounts gambled in those areas prior to the experiment and the week after (Tables 2, 4, and 6). This is further evidence that the results in the area of Odorant No. 1 are not due to
extraprotocol variables, that is, change in gambling characteristics
Table 5. Percent Change in Amounts Gambled on Sunday
Doe to the Presence of Odorant No.1.
Pen:ent
Slot Machine
Pen:ent In<:reru!e When
Odorant Was Added
Over PreviOWJ Sunday
1
2
3
3.37
20.23
57.68
4
9.09
5
6
7
6
9
10
11
12
34.68
-40.54
57.18
3.00
2S.88
52.57
-20.03
67.92
39.56
77.00
81.76
41.50
-49.13
67.39
10.25
-77.89
59.66
81.27
98.14
70.92
-48.21
248.43
52.68
0.59
177.04
194.93
7.96
13
14
15
16
17
HI
Decrease the
Percent Jnereaee Over
Sunday Odorant
Average Sunday
No. 1 Was Removed
AMBIENT ODORS AND SUYr-MACHINE USAGE
Ld
60L9·908· ~ 88
19.20
44.10
70.67
21.05
73.85
-40.53
-69.42
-33.58
-33.12
-101.12
8.72
-57.21
-115.93
-138.84
-70.95
-59.74
57.82
-303.42
-142.48
-377.52
-213..91
-103.74
45.47
591
J86UISU98 J960tj
d5g:~~ o~ OG
uer
Table 6. Percent Change in Amount Gambled on Sunday-Data ADa1yzed
by Student's t Test of Difference of Independent Means.
OdorNo.l
Area
First versus second Sunday
Mean difference
Significance
45.9%
Control
Area
Odor No.2
Area
-1.6%
(NS)
2.6%
(NS)
(NS)
-29.1%
(NS)
2.5%
(NS)
Second vei"SUS third Sunday
Mean differenre
Significance
-35.8%
p<.005
-27.9%
(NSJ
(NS)
Average of first and tbiro Sunday
ver:sus second
Mean difference
Signifu:ance
33.60%
p < .003
13.17%
(NSJ
First versus tbiro Sunday
Mean difference
Significance
p < .03
-6.4%
5.3%
-3.98%
(NS)
of subjects or change of weather, but that Odorant No. 1 did induce
a significant increase in the amounts gambled in the quarter slot
machines surrounding it, whereas Odorant No. 2 made no significant difference in the amounts gambled in the nickel and quarter
slot machines surrounding it.
DISCUSSION
Odorant No. 1 apparently acted to enhance the gambling mood of
the casino patrons in the slot-machine area where it was present. As
to the mechanism of its action, it was not subliminal, because the
amounts used were noticeable (at suprathresbold levels). Nor have
either of the odors, which are GRAS approved (generally recognized
as safe by the U.S. Food and Drug Administration), ever been shown
to impair judgment of nervous functioning, in compliance with
OSHA Hazard Communication No. 29CFR1910.1200 (International
Flavors & Fragrances, 1991). And it seems unlikely that Odorant
No. 1 acted by exacerbating pathological gambling behavior, because
studies of psychiatric effects of odors (Hirsch, 1991) have not yet
shown any effects on obsessive-compulsive disorders.
Odorant No. 1 might have masked unpleasant smells in the area,
but we do not bl!lieve that was the mechanism of action because no
unpleasant odors were noticed in the area prior to the odorization.
Moreover, both Odorant No. 1 and Odorant No. 2 were subjectively
rated as pleasant by an independent panel and both would have had
the same masking effect. It is possible that part of Odorant No. l's
59ll
g·d
HIRSCH
60L9-908- ~ 88
J86UISU98 J960tj
d5g:~~ o~ OG
uer
effect was simply to cause patrons to linger longer; this mechanism
of action has been demonstrated in studies in retail stores that were
odorized (Knasko, 1989). But again, Odorant No. 2' would probably
have had the same effect.
One possible mechanism of action of the effective odorant, we believe, is the phenomena called olfactozy-evoked recall, in which, as
mentioned, memories are stimulated as are their accompanying nostalgic exaltation of mood. Of almost 1000 subjects queried recently,
86% said that an odor could induce in them a vivid memory of the past
(Hinlch, 1992). It seems possible that Odorant No. 1 induced nostalgic
recall and the associated emotions were affectively congruent with,
and enhanced, the gambling mood.
Other studies of odor have emphasized that their effects on behavior
depend upDn their affective congruence with the situatiDnal context.
Mark Twain (Clemens, 1957) dramatically portrays this in "The Invalid's Story" in which a man becomes increasingly distressed while
riding in a f'teight car with a coffin containing a pungent, putrid
corpse. Finally, unable to tolerate the nauseating smell any longer, he
jumps off the train into a winter storm, contracting a fatal illness. But
before dying, he learns that the coffin in the freight car was actually a
large crate of cheese. Hence, an odor, when matched to a certain environment and interpreted in a certain context, can validate intentions
and precipitate action.
However, if individuals do not intend to gamble, it is highly doubtful
that an odor can cause them to. Odorant No. 1 in our study may have
acted somewhat like the bells, whistles, lights, and other environmental stimuli at the casino.
Did Odorant No. 1 merely divert gamblers from surrounding slot
machines, or did it cause those in its area to increase their amount of
gambling? Our results suggest that the latter was the ease. Although
slot machine revenue in the area of Odorant No. 1 increased by
45.11%, revenue in the area of Odorant No.2 and in the control area
did not significantly decrease. This demonstrates that rather than attracting gamblers from other areas, the effect of Odorant No. 1 was
simply to increase the amount of gambling in its own area.
Why Odorant No. 1 had a greater effect on the first day, Saturday
than on the second day, Sunday, may be explained in various ways: The
novel experience of the odor may have been an attractant for gamblers,
but this is doubtful, because the same people were not necessarily present on the first and second days of the experiment. Also, as mentioned,
the novel odor did not attract gamblers away from the unodorized areas.
The most likely explanation relates to odorant levels. On the second
day of the experiment, the level of odorants was significantly below
that of the first day, but still at a suprathreshold level So the lesser
effect on the second day, Sunday, suggests a positive correlation between the level of Odorant No. 1 and the amount gambled. Possibly a
AMBIENT ODORS AND SLQT.MACHJNE USAGE
6'd
60L9-908· ~ 88
593
JS6UISU98 J960tj
d5g:~l
Ol OG uer
higher level of odorization than used on the first day, Saturday, would
have had an even greater impact.
The amounts of money gambled in slot machines in the vicinity of
Odorant No. 2 were not significantly different than the amounts gambled in machines in unodorized control areas. So clearly, not all pleasant
odors will have an impact on the amount of gambling. And it seems
likely that some odors might impact negatively.
REFERENCES
Clemens, S.L. (1957). The invalid's story. In C. Neider (Ed.), Compkte short
stories of Mark Twain. Garden City, NY: International Collectors' Library.
Hirsch, A.R. (1991). Olfaction and psychiatry. Paper presented at the 144th
Annual Meeting, American Psychiatric Association, New Orleans, LA.
Hirsch, A.R. (1992). Nostalgia: A neuropsychiatric understanding. Aduances
in Consu1TU!r Research, 19, 390~395.
Hirsch, A R., &: Gay, S.E. (1991}. Effect on ambient olfactory stimuli on the
evaluation of a common consumer product. Chemical Senses, 16, 535.
International Flavo:rs & Fragrances (1991). IFF material safety data sheet.
Union Beach, NJ:Author.
Knasko, S.C. (1989). Ambient odar and shopping behavior. Paper presented
atAChemS-Xl (Abstract 14.3) Chemical Senses.
Maclean, P.D. (1973). Triune concept of the brain and behavior. 'Thronto: University of Toronto Press.
The author wishes to thank Lee Skelley for his assistance with the project
and Stanley Block for helpful comments on an early draft of this article.
Dr. Hirsch is Instructor, Department of Neurology and Assistant Professor,
Department of Psychiatry, Rush-Presbyterian-St. Luke's Medical Center,
Chicago, IL.
HUISCH
594
o~·d
60L9·908· ~ 88
JS6UISU98 J960tj
eoo:<; ~ o ~
~G
uer