Professional Documents
Culture Documents
DIPAYAN BISWAS
COURTNEY SZOCS
BRIAN WANSINK
ROGER CHACKO *
Dyson Professor of Marketing, Dyson School of Applied Economics and Management, Cornell
and Marketing Officer at Carlson Rezidor Hotel Group. The authors thank the JMR Editor,
ABSTRACT
Retail atmospherics is emerging as a major competitive tool, and it is especially notable in the
restaurant industry where lighting is used to create the overall ambience and influence consumer
experience. Along with influencing overall experience, can ambient light luminance have
unintended consequences in terms of influencing what a diner orders? The results of a field study
at multiple locations of a major restaurant chain and a series of lab studies robustly show that
consumers tend to choose less healthy food options when ambient lighting is dim (vs. bright).
Process evidence suggests that this phenomenon occurs because ambient light luminance
influences mental alertness, which in turn influences food choices. While restaurant and perhaps
grocery store managers can use these insights and their ambient light switches to nudge
consumers toward targeted food choices, such as healthy or high margin signature items, health
conscious consumers can opt for dining environments with bright ambient lighting.
Keywords: ambient light, retail atmospherics, luminance, healthy and unhealthy food choices,
sensory marketing.
3
If a restaurant makes its ambient lighting brighter versus dimmer, would the patrons order
differently? Along with having implications for consumer health and wellbeing, this is a
managerially relevant research question since retail and restaurant managers are increasingly
from competitors (Broniarczyk and Hoyer 2006; Carroll 2012; Weitz and Whitfield 2006). More
broadly, marketers are placing greater emphasis on in-store ambience and displays as strategic
marketing tools (Chandon et al. 2009). Ambient factors can be especially critical since managers
can make subtle and inexpensive changes to the store or restaurant ambience on a regular basis,
While prior research has extensively examined different critical aspects of atmospherics
(Spence et al. 2014), such as scent (Morrin and Ratneshwar 2003), music (Baker et al. 2002;
Bruner 1990), flooring (Meyers-Levy, Zhu, and Jiang 2010), and ceiling height (Meyers-Levy
and Zhu 2007), little is known about how ambient lighting levels might influence specific
product choices. This is especially noteworthy since in many retail/restaurant contexts, managers
can control the ambient light luminance (brightness or dimness) with relative ease and with
Accordingly, this research examines how increasing versus decreasing the luminance of
ambient light might influence choices between food items that are considered healthy versus
unhealthy. Our findings have both conceptual and practical implications. From a conceptual
perspective, the findings of our studies robustly document how ambient light luminance impacts
choices between healthy and unhealthy food options. They thus contribute to the findings of a
wider set of studies, examining choices for healthy/unhealthy options, focusing on such factors
as mode of decision-making (Shiv and Fedorikhin 1999), temptation (Dhar and Wertenbroch
4
2012), self-control (Baumeister 2002; Kivetz and Simonson 2002), display patterns (Romero and
Biswas 2016), and health claims (Chandon and Wansink 2007), among others. We contribute to
this literature stream by examining the role of atmospherics, such as in the form of ambient light
behavior (such as quantity of food consumed), we build our conceptual framework and then test
our hypotheses through five experiments (with one conducted at several locations of a chain
restaurant and four conducted in the lab) along with another experiment reported in the Web
Appendix. We first examine the main effects of ambient light on product choices through a field
experiment (Study 1a) as well as through controlled lab experiments (Studies 1b and 1c) and find
that preference for healthy food options is higher when choices are made in bright (vs. dim)
ambient light luminance. Studies 2a-2b provide process evidence for the proposed theorization
by examining the moderating effect of inducing alertness. We also report the results of an
additional study, in the Web Appendix, where we examine the moderating effects of inducing
CONCEPTUAL BACKGROUND
Ambient lighting can influence stimulation levels, cognitive associations, and overall
behavior in general (Spence et al. 2014) and hence can be a potentially effective tool for
marketers. Moreover, since it is easy and inexpensive to alter ambient lighting, not surprisingly,
stores and restaurants vary greatly in terms of ambient light luminance. Light luminance is
measured in lux (lumens per square meter) (Thimijan and Heins 1983). To put things in
5
perspective, the luminance of full outdoor daylight is approximately 10,000 lux, twilight is
approximately 11 lux, the recommended luminance level for normal office work is 500 lux, and
medical facilities and operating rooms are usually required to have luminance levels above 1,000
lux (Bosworth Instruments 2016; Nielsen, Svendsen, and Jensen 2011; The Engineering Toolbox
In the context of restaurants, based on anecdotal evidence, dimly lit restaurants tend to
have ambient light luminance between 10 and 40 lux; however, some restaurants have more
extreme levels of ambient light luminance. For example, Qi, a restaurant in New York City (near
Times Square) has such dim ambient lighting that many people find it difficult to read the menu
inside the restaurant. Along similar lines, the menus at Cero’s Speakeasy (a restaurant in Tampa,
Florida) come with reading lights attached since the ambient lighting is so dim that diners could
otherwise not see the menu. At an even more extreme level, it is literally pitch dark at the
Opaque restaurant chain and the company capitalizes on the idea of dark dining as a point of
Incidentally, there are norms and expectations associated with lighting and other aspects
of atmospherics; for example, fast food restaurants tend to have bright lights and fast music
while fine dining restaurants tend to have dim lights and relaxing music (Wansink and van
Ittersum 2012). There might also be self-selection issues, whereby certain consumers might
prefer restaurants with dimmer versus brighter lighting (Spence et al. 2014).
The limited research examining the effects of ambient light has had a much different
focus than on choice. For instance, Wansink and van Ittersum (2012), in an experimental study at
a Hardee’s fast food franchise restaurant, softened the lighting and music in the restaurant to
create a relatively fine dining experience condition and found that consumers spent more time
6
and ate less in the “fine dining” area than in the main dining area. Since these consumers were
assigned to the lighting (bright or dim) condition after they had selected their food, it is not
known if such a change would have influenced initial food choices. Along the same lines,
research shows that soft or warm lighting tends to cause people to stay longer and enjoy an
unplanned dessert or drink (Wansink 2004). However, from these studies, it is unclear as to how
lighting might influence the healthfulness of the ordered foods and beverages.
Wansink et al. 2012) manipulated visual cue availability of food items by having diners eat in
either the presence versus absence of light. They found that consumption volume was higher
when participants ate in the absence of light (in total darkness) than in the presence of light. In
these studies, all participants ate the same food items and hence it is again not clear how ambient
Prior studies have also examined the effects of light on consumers’ wall color perceptions
and subsequent product evaluations. For instance, Oberfeld et al. (2009) found that white wine
tasted better in rooms where lighting was used to make the walls seem blue or red. Similarly,
Areni and Kim (1994) found that bright versus soft ambient light led to consumers handling and
examining more bottles of wine, with the effects being moderated by shelf level of the wine; it is
not clear though what the light luminance levels were across the conditions or how lighting
influenced choices.
Although these prior studies demonstrate interesting findings, none of these studies
focused on ambient light luminance level and its potential effects on choices involving healthy
and unhealthy items. As a result, it remains unclear how ambient light luminance level might
influence consumers’ choices between healthy versus unhealthy food options. Moreover, our
7
studies differ from these prior studies because we do not examine the absence versus presence of
light (i.e., total darkness vs. presence of lighting) nor do we examine the effects of light color
influencing wall color perceptions. Instead, in our studies, across all conditions, light is present
and is of the same color but varies only in the level of luminance (i.e., brightness/dimness).
We next discuss factors that can influence choices between relatively healthy and
unhealthy foods. Following this, there is a discussion of how ambient light might influence
product choices by affecting decision making modes, alertness levels, inhibition, and focus on
self-presentation.
When choosing between healthy and unhealthy options, consumers are in essence
choosing between options that dominate on affective/hedonic dimensions and thus appeal to the
heart (i.e., unhealthy options) and options that dominate on cognitive/utilitarian dimensions and
thus appeal to the mind (i.e., healthy options) (Khan, Dhar, and Wertenbroch 2005; Shiv and
Fedorikhin 1999). When faced with conflicts of the heart and mind, the consumer’s decision-
making mode plays an influential role in determining the choice outcome. Specifically, when
choices are made in a more deliberate (or cognitive) manner, there is increased preference for
virtues or healthy food options (Shiv and Fedorikhin 1999). As a result, there tends to be a
positive correlation between the degree of attentional resources devoted to the choice task and
preference for the healthy option (Dhar and Wertenbroch 2012). Along similar lines, research
shows that mindful eating leads to healthier food consumption (Wansink 2006). Moreover, at a
broader level, choices between healthy and unhealthy options are often influenced by trade-offs
health/wellbeing (Romero and Biswas 2016). Greater degree of deliberate and cognitive
processing tends to enhance focus on long-term benefits (Dhar and Wertenbroch 2012; Gardner
et al. 2014).
So between high and low levels of ambient light luminance, which condition would
affective decision-making? Research suggests that dim (vs. bright) ambient light reduces mental
alertness (Cajochen 2007), inhibition (Hirsch, Galinsky, and Zhong 2011), and self-presentation
focus (Kasof 2002; Zhong, Bohns, and Gino 2010). As will be discussed in detail below, these
theoretical accounts would predict greater preference for healthy options in bright (vs. dim)
ambient light.
Sleep research as well as ergonomics research show that bright lighting enhances mental
alertness and task performance because among other physiological changes, bright light
suppresses melatonin, which is the primary controller of circadian (day/night) sleep bio-rhythms
(Crowley et al. 2003; Daurat et al.1993; Lowden, Åkerstedt, and Wibom 2004). In contrast to the
alerting effects of bright lighting, low levels of light luminance (i.e., dim lights) have been
shown to increase sleepiness and reduce alertness (Badia et al. 1991; Lowden, Åkerstedt, and
Wibom 2004).
production, core body temperature, heart rate, and cortisol production, all of which are correlated
with alertness levels (Lockley et al. 2006). Specifically, bright (vs. dim) light influences human
and an increase in cortisol levels, along with physiological changes in terms of enhanced core
body temperature, and also psychological changes in the form of reduced sleepiness and
In the context of the present research, these literature streams suggest that consumers will
be less mentally alert in dimly (vs. brightly) lit environments due to psychophysiological factors.
Additional research across different domains has established a positive relationship between
mental alertness, attention, and cognitive performance (Lim and Dinges 2008; Thomas et al.
2000). In fact, emerging research in neuroscience shows that ambient light can modulate cortical
activity related to alertness, which in turn can stimulate cognitive functions and activity (Virginie
et al. 2015). In essence, a higher level of mental alertness enhances cognitive performance.
Therefore, it can be proposed that in dim (vs. bright) ambient light, consumers will be less alert
mentally and hence exert reduced cognitive processing when choosing between the healthy and
When consumers choose between healthy and unhealthy food items, a lower level of
cognitive availability tends to lead to greater preference for unhealthy options (Dhar and
Wertenbroch 2012; Shiv and Fedorikhin 1999). Moreover, research also shows that reduced
mental alertness leads to mindless decisions (Janssen et al. 2008) and mindless decisions tend to
lead to unhealthy behavior (Wansink, Just, and Payne 2009). Thus, based on our preceding
discussions, we propose that in dim (vs. bright) lighting there will be greater preference for the
unhealthy food item. In other words, there will be greater preference for the healthy option when
H1: When given a choice between healthy and unhealthy options, consumers will have
greater preference for healthy options when ambient light luminance is bright (vs. dim).
10
Theories related to self-presentation focus and inhibition also predict similar behavioral
outcomes to that predicted by H1. Research in the domain of self-presentation focus suggests
that focus and concern about self-presentation is enhanced in the presence of bright light (Kasof
2002; Zhong, Bohns, and Gino 2010). Since enhanced focus on self-presentation is likely to lead
to greater degree of preference for healthy options, there should be higher choice likelihood of
the healthy option when ambient light is bright (vs. dim), consistent with H1’s prediction.
In a related vein, research in the domain of inhibition suggests that consumer decision-
making tend to be more disinhibited in dim (vs. bright) ambient light. This is mainly because dim
lighting gives people a sense of perceived anonymity, which in turn encourages moral
transgressions (Zhong, Bohns, and Gino 2010). In the context of choosing between healthy and
unhealthy product options, disinhibition is likely to enhance preference for the unhealthy (vs.
healthy) option since enhanced disinhibition would imply reduced self-control, which in turn
would reduce the likelihood of choosing the healthy option (Romero and Biswas 2016). In other
words, theories related to inhibition would also suggest that preference for the healthy option
should be higher when ambient light is bright (vs. dim). In essence, research on self-presentation
mental alertness and disinhibition are different constructs, they can have a causal (Baumeister,
Heatherton, and Tice 1994; Herman and Polivy 1993) or correlational relationship (Verwey and
Zaidel 2000). That is, reduced mental alertness can lead to disinhibition (Baumeister, Heatherton,
and Tice 1994; Herman and Polivy 1993). Similarly, extraneous factors can affect both these
constructs; for instance, a high level of intoxication not only leads to reduced mental alertness,
11
but also reduces inhibition (Heinz et al. 2011; Steele and Josephs 1990). Along similar lines,
make people more inhibited (Heatherton, Striepe, and Wittenberg 1998; Leary and Atherton
1986).
focus, H1 predicts higher preference for the healthier option when ambient light is bright (vs.
CHAINS
Method
chain, in collaboration with the corporate management. The restaurants where the study was
conducted are part of a major casual dining chain with over 1200 locations in 23 countries. The
four locations where the study was run are all based in the same metropolitan area of a major city
in the US, and these locations are in close proximity to each other. This study was a single factor
The experiment was conducted at the four restaurant outlets on a single random weekday.
The study was undertaken on a single day to avoid potential confounds related to weather and
other extraneous variables. The study was run between 6 and 8 pm. Two of the restaurants,
randomly chosen, had dim lighting while the other two had bright lighting. In the dim ambient
light condition, the restaurants had a low light luminance level of 25 lux while in the bright
12
ambient light condition, the light luminance was set at 250 lux. The light luminance levels were
based on a pretest using interactive feedback from the restaurant staff (at a different location than
where the main studies were conducted). Specifically, the restaurant staff at the pretest site felt it
was too dark below 25 lux and anything above 250 lux was deemed as too bright for the
restaurant. The usual luminance level at these restaurants varied between 100 and 125 lux and all
To avoid implicit demand effects, none of the researchers were involved with selecting or
patrons and ask them to fill out a short survey. The staff members at these locations did not have
any information about the hypotheses or purposes of the study. Senior executives from the
corporate office oversaw the procedure to ensure consistency across the four locations. Each
restaurant was instructed to get approximately forty completed surveys. This led to a total of one
hundred sixty restaurant patrons (51% females), completing the survey (N = 78 for the dim light
restaurants and N = 82 for the bright light restaurants), across the four locations.
The study survey asked participants which menu item(s) they ordered and the restaurant
staff unobtrusively verified whether the item indicated on the survey matched the actual order.
Participants were also asked about their alertness level, with a reverse-coded measure (1 = very
alert, 7 = not at all alert). Patrons also indicated their age in brackets, with 19% being between
21-29, 16% between 30-39, 19% between 40-49, 23% between 50-59, 17% between 60-69, and
6% at 70+.
The researchers, a priori, coded each item on the restaurant menu as healthy versus
unhealthy based on prior commercial standards (Pope et al. 2014). Specifically, grilled and baked
fish, white meat (chicken and turkey), and vegetables were coded as healthy while fried food
13
items and red meat (beef and pork) were coded as unhealthy. Each restaurant patron’s order was
irrespective of whether they ordered something healthy along with it, the overall meal was coded
as “unhealthy” ordering (e.g., Chernev 2011). For example, if someone ordered a steak and a
salad, it was coded as “unhealthy” ordering. We also computed the calorie content for the
Results
Main tests for choice and alertness. Consistent with H1, for the restaurant locations with
bright (vs. dim) ambient light, a higher proportion of patrons ordered healthy foods (52.44% vs.
34.62%; χ2 = 5.16, p < .05). Figure 1 graphically presents the key findings. Also, consistent with
our theorizing, mental alertness was higher in the bright (vs. dim) ambient light condition (5.28
Since females (vs. males) tend to choose healthier options (Wardle et al. 2004), we ran
the analysis with gender as a covariate to rule out the alternative explanation of the effects being
driven by gender distribution differences across the conditions. Including gender as a covariate
made the effects stronger (χ2 = 5.66, p < .02). We also examined the effects of gender, as an
independent variable, on food choices. While gender had directional main effects on food
choices, with females (vs. males) directionally choosing healthier options to a greater extent, the
effects were not significant (49.35% vs. 37.84%; χ2 = 2.03, p = .15). There was no interaction
effect between gender and ambient light on food choice (p > .45).
Main tests for calories. We also analyzed the data with total calories ordered, as a
continuous variable. There was a significant effect of ambient light on total calories purchased,
14
with calorie purchases being higher for dim (vs. bright) lighting (Mdim-light = 1336.49, SD =
598.36 vs. Mbright-light = 962.56, SD = 580.70; F(1, 158) = 16.09, p < .001). Overall, customers in
dim ambient light settings purchased 38.85% more calories than those in bright ambient light
settings.
Mediation tests. Mediation analysis using Preacher and Hayes’s (2008) PROCESS macro
Model 4 with 5,000 bootstrapped samples (Hayes 2012) shows indirect effects of ambient light
luminance on food choices, with the effects being mediated by mental alertness (B = -.226, SE =
.136, 95% CI95: -.571, -.029), as evidenced by the CI (confidence interval) excluding zero.
Similar indirect mediation effects emerged for calories ordered (B = -52.106, SE = 32.229, 95%
Discussion
The results of Study 1a demonstrate that consistent with our hypothesis, restaurant
patrons tend to order healthier items and fewer calories when dining in a brightly (vs. dimly) lit
restaurant. Process evidence shows that mental alertness mediates the effects of ambient light on
food choices and calories purchased. Specifically, these results support the theoretical premise of
mental alertness being higher in bright lighting, which in turn leads to higher degree of ordering
of healthy options, and lower level of calories purchased, in bright (vs. dim) ambient lighting. It
should be noted though that since alertness was measured after customers placed their orders,
this might be a correlational effect instead of a causal effect. Since Study 1a was conducted
across four different locations, there might have been potential differences across these locations,
15
such as in clientele profiles that could have had confounding effects. Hence, next, Study 1b
Method
Study 1b was a single factor between-subjects design experiment with two manipulated
conditions (ambient light luminance: bright vs. dim). One hundred thirty university students
(average age 23 years, 53.5% females) participated in the experiment in exchange for course
credit. It should be noted that across all our studies, the key dependent variable is food choice
and the average sample size for each condition is over 35 participants; this target sample size was
determined based on the approach adopted in prior research with choice (a dichotomous
variable) as the DV (e.g., White et al. 2016). Also, across all our studies, the entire data was
collected in one round and all the analyses were conducted only after all the data were collected.
In Study 1b, all the participants responded to the question on food choice but two of the
participants did not respond to some of the other questions, including the questions on inhibition
and the alertness task; these two participants were retained in the sample. The bright and dim
lighting conditions had 50.0% and 57.38% females, respectively. The experiment was conducted
in a laboratory with technological options to have any level of ambient light luminance between
In order to ensure ecological validity, in the dim ambient light condition, the lab had a
luminance of 10 lux while in the bright ambient light condition, the luminance was set at 900
lux. Thus, the manipulations of dim and bright lighting conditions were more extreme in this
study than in Study 1a. This is because, unlike field studies, which have managerial constraints,
16
lab studies allow greater flexibility in the experimental manipulations. Importantly, based on
area in the US, a couple of very dim restaurants in the metropolitan area had luminance levels
around 10 lux while a couple of very brightly lit restaurants had luminance levels of around 900
lux. Hence, the chosen luminance levels of 10 lux and 900 lux for Study 1b have ecological
validity.
For the experiment, participants first arrived at a waiting area and were then brought into
the main lab by a research assistant. After participants entered the lab, they were seated at a
table. In the initial few minutes, participants were asked to settle down and then asked if they
needed any pens/pencils, to switch off their cell phones, and also to be quiet during the entire
duration of the study; they were then handed a survey. After this, they were asked to choose
between two food options – 100-calorie Oreos and chocolate covered Oreos. These food items
have been used in prior studies to represent healthy and unhealthy options respectively (Wilcox
et al. 2009). Participants were told they could have only one of these food options and they were
told to record their preference on the survey. We also measured inhibition by asking participants
to indicate their level of agreement (1 = strongly disagree, 7 = strongly agree) with a statement
on behavioral inhibition (“right now, I feel inhibited”) (Duke and Bègue 2015).
Study 1a had self-reported measures of alertness, which provided mediation evidence for
the underlying process. In Study 1b, we attempted to examine the effects of ambient light on
objective measures of mental alertness. Specifically, we measured mental alertness level at the
beginning of the study by using a digit span task, employed in prior research (Irmak, Block, and
Fitzsimons 2005). That is, we assessed participants’ mental alertness levels by showing them a
series of numbers (between 1 and 99) displayed on a screen. Each number sequence was
17
automatically timed to be displayed for exactly half a second, following which, participants were
asked to reproduce the numbers on the surveys, in the exact order in which they were displayed
on the screen. Four such sets were displayed, with the first set having a sequence of five
numbers, progressively going up to seven numbers in the third and fourth sets. In essence, a total
of 25 numbers were displayed across the four sets; see Web Appendix A for details.
Main tests. Consistent with the findings observed in Study 1a, a higher proportion of
participants preferred the healthy option with bright (vs. dim) ambient light (68.12% vs. 49.18%;
χ2 = 4.74, p < .05). The results of Study 1b again support H1. We also ran the analysis with
gender as a covariate. The results of this analysis show that having gender as a covariate leads to
equivalent result patterns for main effects of ambient light on food choice (χ2 = 4.67, p < .05).
We also examined the direct effects of gender on food choice. There was a main effect of gender
on food choice, with females (vs. males) choosing healthier options to a greater extent (66.18%
vs. 49.15%; χ2 = 3.77, p < .06). There was no interaction effect between ambient light and gender
The self-reported measure of behavioral inhibition was similar across the two ambient
lighting levels (5.60 vs. 5.46; F(1, 126) = .32, p = .57), which suggests that disinhibition was not
mental alertness, which in turn would influence choices between healthy versus unhealthy
options. The digit span task, although unrelated to the food choice task, gave an objective
measure of alertness level. The number of correct responses from the digit span task was
18
recorded. Results from the digit span task show that participants were able to observe and
reproduce the numbers (that were displayed for half a second) with greater accuracy when they
were in bright (vs. dim) ambient light (Mbright-light = 10.94 vs. Mdim-light = 9.98; F(1, 126) = 4.88, p
< .05).
Study 1b replicated the key findings of Study 1a in a controlled lab setting, and also
provided additional evidence for our theorizing. Specifically, while Study 1a had self-reported
measure of alertness, Study 1b tested our theoretical claims with an objective measure of
alertness. Next, Study 1c examines if the effects of Study 1b hold when participants are made to
Method
Study 1c was similar to Study 1b in terms of procedure, with the only difference being
that while in Study 1b participants anonymously recorded their food preference on a paper
survey, in Study 1c participants also had to indicate their choices aloud. The objective behind
this approach was to examine if eliminating the perceived anonymity of choices might change
the pattern of results observed in Study 1b. In a group setting, indicating preferences aloud
(instead of just recording preferences on a survey) draws attention to the self, which in turn
(Fern 1982).
To enhance the robustness of the findings across studies, a different set of food options,
chocolate versus granola bar (Laran 2010), was used in this study. Similar to the approach used
in Study 1b, participants first arrived at a waiting area and were then randomly assigned to a
19
session and the ambient lighting for each session was also randomly determined. In the lab, all
After the initial few minutes (similar to the procedure used in Study 1b), participants
were told that there were two food items, a chocolate bar and a granola bar, and that that they
could choose only one of these for eating. A sample of each food item was displayed at the
center of the table. Participants were told to indicate their response on the survey and then raise
their hand and indicate their preference aloud to the researcher. A research assistant then brought
the participant her/his preferred item. Seventy-one university students (average age 22 years,
47.7% females) participated in the experiment in exchange for course credit. In the bright and
dim light conditions, there were 41.2% and 58.4% females, respectively.
Consistent with the findings observed in Studies 1a-1b and as predicted by H1, there was
greater preference for the healthy option with bright (vs. dim) ambient light (54.29% vs. 30.56%;
χ2 = 4.10, p < .05). Including gender as a covariate enhanced the effect (χ2 = 6.26, p < .02). There
were no main effects of gender on food choice (χ2 = .32, p = .57) and neither were there
interaction effects between gender and lighting on food choice (p > .70).
Study 1c had two objectives. First, it replicated the effects of Studies 1a and 1b with a
different set of product options. Second, the results of Study 1c again highlight how mental
determining the outcomes of Studies 1a-1b. That is, if inhibition had a more dominant role in
determining the food choice outcome, then the pattern of results should have been different when
participants had to indicate their choice aloud. However, indicating choice aloud led to the same
20
anonymously on a survey. Although this does not technically rule out the role of inhibition as an
underlying process, it does highlight that factors that enhance inhibition levels do not necessarily
change the pattern of results related to effects of ambient light on food choices. Next, Studies 2a-
2b provide additional evidence for the process driving the effects of ambient light on food
The results of Studies 1a-1c support a mental alertness based explanation for the effects
of ambient light on product choices; Studies 2a and 2b provide more direct evidence for this
proposed theorization. Specifically, we theorized that mental alertness would be lower when
ambient light is dim (vs. bright), which in turn will lead to unhealthier choices under dim (vs.
bright) ambient light. If our conceptualization holds, then this differential effect of dim versus
bright ambient light on food choices would get attenuated when consumers’ mental alertness
levels are enhanced. That is, we propose that the effects predicted by H1 and demonstrated in
Studies 1a-1c, would hold under regular mental alertness levels but will get attenuated when
H2: Under regular mental alertness levels, when given a choice between healthy and
unhealthy options, consumers will have greater preference for healthy options when
ambient light luminance is bright (vs. dim).
Under enhanced mental alertness, these effects will get attenuated, whereby the
preference pattern will be the same across bright and dim ambient lights.
Method
21
H2 was tested in Study 2a with the help of a 2 (ambient light luminance: bright vs. dim)
X 2 (mental alertness level: regular vs. high) between-subjects experiment. The procedure was
similar to Studies 1b-1c whereby participants first arrived at a waiting area and were then
brought into a lab and seated at a table. The ambient light luminance in the lab was manipulated
in the same manner as in Studies 1b-1c (i.e., 10 lux for dim ambient light and 900 lux for bright
ambient light). Mental alertness level was manipulated through placebo effects, associated with
sampling a beverage, consistent with the approach adopted in prior studies (Biswas, Grewal, and
Roggeveen 2010; Shiv, Carmon, and Ariely 2005). Participants were first given a beverage,
which was identical across all conditions. In the “high alertness” conditions, participants were
told that the beverage contained a high level of caffeine. No such statement was provided in the
“regular alertness” conditions. A pretest (N = 34; mean age 25 years; 50% females) was
conducted to see if providing such a statement did indeed enhance mental alertness. The results
of the pretest showed that participants reported that they felt more alert (measured on a 1-7 scale,
where 1 = low alertness and 7 = high alertness) after sampling the beverage when they were told
it contained caffeine versus when they were not (Mcaffeine-present = 5.67 vs. Mcaffeine-absent = 4.26;
In the main study, participants were asked to drink the beverage, and to avoid hypothesis
guessing, they were asked a couple of questions about the taste of the beverage. After this,
participants completed the focal task, which involved choosing between a healthy (baked potato)
and an unhealthy (fries) food item (Wilcox et al. 2009). They were also asked how mentally alert
they felt at that point (with 1 = not at all alert, 7 = extremely alert).
Three hundred fifty three university students (average age 23 years; 49.6% females)
participated in this experiment in exchange for course credit. In the “caffeinated” (i.e., high
22
alertness level) condition, there was equivalent proportion of females in the bright and dim light
conditions (40.30% and 45.78%) and there was a similar pattern for the “non-caffeinated” (i.e.,
Results
Main tests. The results of a 2 (ambient light luminance: bright vs. dim) X 2 (mental
alertness level: regular vs. high) logistic regression revealed a significant interaction effect on
food choice (Wald χ2 = 6.34, p < .05). Including gender as a covariate demonstrated a similar
level of interaction effect (Wald χ2 = 6.56, p < .05). There was no main effect of gender on food
choice (χ2 = .62, p = .43) and there was no interaction effect between gender and ambient light
Follow-up tests showed that consistent with H2, when no caffeine information was given,
choice for the healthy item was higher with bright (vs. dim) ambient light (49.0% vs. 32.67%; χ2
= 5.55, p < .05) but the effects got attenuated when participants were given a “caffeinated”
beverage (46.27% vs. 54.12%; χ2 = .92, p = .34). It is interesting to note that consistent with our
theorizing, with bright ambient light, the choice pattern for healthy versus unhealthy items
remained the same irrespective of whether the alertness level was regular versus high (49.0% vs.
46.27%; χ2 = .12, p = .73); with dim ambient light, the preference for healthy food was enhanced
when mental alertness was high (vs. regular) (54.12% vs. 32.67%; χ2 = 8.69, p < .01). These
findings are consistent with our conceptualization and predictions. That is, in bright ambient
lighting, mental alertness level was high by default and hence the “caffeinated” beverage did not
have any effect. In contrast, in dim ambient lighting, mental alertness level was lower and the
23
“caffeinated” beverage helped in enhancing the alertness level and thus the preference of the
healthy item.
Test of moderated mediation. Moderated mediation analysis using Preacher and Hayes’s
(2008) PROCESS macro Model 8 with 5,000 bootstrapped samples (Hayes 2013) shows
moderated mediation of alertness for the interaction effect between ambient light and alertness
level on food choice (B = .193, SE = .109, CI95: .030, .464). Consistent with our expectations, for
the “regular alertness” condition, the indirect effects of ambient light on food choice was
significantly mediated by alertness (B = .144, SE = .076, CI95: .028, .329); there was no such
significant mediation effect for the “high alertness” condition (B = -.049, SE = .068, CI95: -.206,
.066). These results again highlight the role of alertness as the underlying process for the effects
Ruling out alternative explanations. As mentioned earlier, alternative explanations for the
observed effects can be provided by theories related to inhibition and self-presentation focus,
which predict the same behavioral outcomes for dim versus bright ambient lighting as mental
alertness theories. While Study 1b examined the effects on self-reported measures of inhibition,
examine the role of self-presentation focus, participants were asked to indicate their level of
agreement with three statements (adapted from Fenigstein, Scheier, and Buss 1975) such as
“Right now, I am very concerned about the way I am presenting myself,” “Right now, I am
worried about making a good impression,” and “Right now, I am aware of what other people
think of me” (1=strongly disagree, 7=strongly agree). One participant did not respond to the self-
showed a non-significant interaction effect on self-presentation focus (F(1, 348) = 1.55, p = .21).
24
Follow-up tests showed that self-presentation focus level was similar across bright (vs. dim)
ambient lighting (Mbright-light = 3.33 vs. Mdim-light = 3.14; F(1, 350) = 1.13, p = .29). These results
provide further support for the contention that mental alertness, as opposed to self-presentation
focus, is the dominant underlying process for the effects observed in our studies.
related to healthy/unhealthy lifestyle. Accordingly, towards the end of the survey, we measured
health orientation by asking participants to indicate the extent to which they disagreed/agreed (1
= strongly disagree and 7 = strongly agree) with two items: “Calorie levels influence what I eat”
and “Eating healthy is important to me” (Chandon and Wansink 2007). Four participants did not
respond to the health orientation questions. The mean score on these two items were similar
across the “caffeinated” and “non- caffeinated” conditions, both in the bright light (4.75 vs. 4.44;
F(1, 345) = 1.79, p = .18) and dim light (4.82 vs. 4.63; F(1, 345) = .77, p = .38) conditions.
Discussion
influencing the effects of ambient light luminance on choices between healthy and unhealthy
options. Under regular mental alertness, similar to the effects observed in Studies 1a-1c, there is
greater preference for the unhealthy option when ambient light was dim (vs. bright). However,
when mental alertness is enhanced through a placebo effect, the effects get attenuated. While
enhancing mental alertness did not have any effect in the case of bright ambient light, it did
influence choices in the case of dim ambient light. Specifically, with dim ambient light, there
was greater preference for the healthy option when mental alertness was enhanced. Tests of
While Study 1b examined the role of inhibition level, Study 2a examined the potential
effects of self-presentation focus and also the moderated mediation effects of alertness. The
results of both these studies suggest that although inhibition and self-presentation make similar
prediction as H1, mental alertness seems to be the dominant underlying process. Next, Study 2b
Method
Study 2b was a 2 (ambient light luminance: bright vs. dim) X 2 (induced mental alertness
level: regular vs. enhanced) between-subjects experiment. While in Study 2a, mental alertness
level was manipulated through a placebo effect, Study 2b manipulated mental alertness in a more
direct manner. Specifically, in the “enhanced mental alertness” conditions, mental alertness was
manipulated by asking participants, at the beginning of the study, to be mentally alert while
undertaking the tasks in the study. No such instructions were given in the “regular mental
alertness” conditions.
Continuing with the objective of having different food choice scenarios across the
studies, mainly to ensure robustness, in Study 2b, participants had the option of choosing
between raisins and M&Ms (Salerno, Laran, and Janiszewski 2014). Mental alertness was
measured by asking participants: “How alert do you feel right now?” (1 = not at all alert, 7 =
very alert). One hundred forty nine university students (average age 22 years; 53% females)
participated in this experiment for course credit. In the bright and dim lighting conditions, 51.3%
and 54.8% were female, respectively. The procedure, with the exception of the mental alertness
Results
Food choice. The results of a 2 (ambient light luminance) X 2 (mental alertness level)
logistic regression revealed a significant interaction effect on product choice (Wald χ2 = 4.73, p <
.05). Including gender as a covariate keeps the overall effect and chi-square value unchanged (at
χ2 = 4.73). There was no main effect of gender on food choice (χ2 = 1.40, p = .24) and there was
no interaction effect between gender and ambient light on food choice (p > .50).
Follow-up tests showed that choice of the healthy item was higher with bright (vs. dim)
ambient light when there was no “alertness” inducement (27.45% vs. 7.50%; χ2 = 5.87, p < .05)
but the effects got attenuated when mental alertness was enhanced (28.0% vs. 30.30%; χ2 = .04,
p = .85). It is interesting to note that consistent with our theorizing, with bright ambient light, the
choice pattern for healthy versus unhealthy items remained the same irrespective of whether the
induced alertness level was regular versus high (27.45% vs. 28.0%; χ2 = .003, p = .96); with dim
ambient light, the preference for healthy food was enhanced when mental alertness was induced
(vs. regular) (30.30% vs. 7.50%; χ2 = 6.42, p < .05). These findings are consistent with our
conceptualization. That is, for bright ambient lighting, mental alertness level was high by default
and hence enhancing alertness did not have any effect. In contrast, for dim ambient lighting,
Alertness effects. The results of a 2 (ambient light luminance) X 2 (mental alertness level)
ANOVA revealed a significant interaction effect on self-reported alertness level (F(1, 145) =
4.79, p < .05). Follow-up tests show that inducing (vs. not inducing) alertness led to higher
perceived alertness level (5.52 vs. 4.23; F(1, 145) = 30.86, p < .01). Also, consistent with our
theorization, self-reported mental alertness level was higher for bright (vs. dim) ambient light in
27
the absence of induced alertness (4.67 vs. 3.68; F (1, 145) = 10.85, p < .01) with the effects
getting attenuated when alertness was induced (5.48 vs. 5.55; F(1, 145) = .03, p = .86).
Discussion
The results of Study 2b replicate the findings of Study 2a using a different type of mental
alertness manipulation and again highlight the role of mental alertness as the dominant
underlying factor in influencing the effects of dim versus bright ambient lighting on food
choices.
GENERAL DISCUSSION
The results of five experiments (one conducted as a field study at multiple locations of a
major restaurant chain and four in labs) show that consumers have greater preference for
unhealthy options when the luminance of ambient light is dim (vs. bright). Similarly, as the
results of Study 1a show, dim (vs. bright) ambient lighting also leads to higher volume of calorie
consumption. Process evidence suggests that dim (vs. bright) ambient light reduces the level of
mental alertness, which in turn leads to greater preference for unhealthy options. These results
contribute to the growing literature on choices between healthy and unhealthy options and also to
the literature on how ambient factors influence food choices. While prior studies have examined
different factors that can influence choices between healthy and unhealthy product options
(Chandon and Wansink 2007; Dhar and Wertenbroch 2012; Romero and Biswas 2016; Shiv and
Fedorikhin 1999), the present research is the first to examine how ambient light influences such
choices.
28
It might be noted though that several factors influence choices between healthy/unhealthy
options and ambient light luminance is only one such factor that can nudge consumers towards
more healthful choices. It also needs to be emphasized that we are claiming relative effects of
ambient light instead of absolute effects. In other words, changing the ambient light can lead to
relatively higher/lower levels of unhealthy choices; however, depending on context and food
options, the overall choice pattern might still be unhealthy. In fact, the results of our Study 2b
demonstrate such a pattern, whereby bright lighting leads to only 27.45% healthy choices;
however, that is still significantly better than dim lighting, which had 7.50% healthy choices.
Even in our field experiment at the restaurants (Study 1a), bright ambient light leads to 52.44%
healthy choices, but it is significantly worse at 34.62% with dim ambient lighting.
The findings of our research contribute to the literature on choice construction in general.
Prior studies examining choice and preference construction have often focused on the role of
product attributes (Amir and Levav 2008) and there have been only a limited number of studies
examining the role of ambient factors in influencing choice construction. The present research
takes an important step in highlighting the role of ambient factors in influencing product choices.
The findings of this research also contribute to the literature on ambient light. While prior
Akerstedt, and Wibom 2004), psychological (Schaller, Park, and Mueller 2003), and behavioral
responses (Zhong, Bohns, and Gino 2010), this is the first study to link ambient light luminance
with consumer choices for healthy and unhealthy options. In addition, the findings of this
research contribute to the growing literature on the role of visual cues and availability of light or
the softening of light on food consumption (Scheibehenne, Todd, and Wansink 2010; Wansink
2004; Wansink et al. 2012; Wansink and van Ittersum 2012). These studies found for example,
29
that softening the light in conjunction with the music led to stronger perceptions of a finer dining
experience and lower consumption volume (Wansink and van Ittersum 2012), and that the
absence of light led to a greater amount of food consumption due to the non-availability of visual
cues (Scheibehenne, Todd, and Wansink 2010; Wansink et al. 2012). We extend this literature
stream by demonstrating how dimming of ambient light enhances choice likelihood for
unhealthy options.
The present research is also possibly the first study to examine the interaction effects
between ambient light and placebo effects (or even induced factors) related to enhancing mental
alertness. As the results of Studies 2a-2b show, factors that can enhance consumer mental
alertness can diminish the effects of ambient light on food choices. We also conducted an
additional experiment (reported in details in the Web Appendix) where we examined the
moderating effects of inducing sleepiness through a priming task and examined choice from a
menu of food options. The results of this study show that inducing sleepiness (which is
conceptually similar to reducing mental alertness) attenuates the effects observed in Studies 1a-
1c, whereby higher level of sleepiness enhances choice likelihood of unhealthy options,
irrespective of ambient light luminance level. Other factors that can potentially influence mental
alertness can be certain types of alerting and sleep-inducing ambient scents (Spangenberg,
Crowley, and Henderson 1996) and music with certain tempos and volumes (Mattila and Wirtz
2001). Clearly, additional work is needed to examine how these other ambient factors (scent and
music) might interact with ambient light and influence product choices.
This research also has important potential implications for sensory marketing and retail
atmospherics. While prior studies in the domains of sensory marketing and retail atmospherics
have examined different types of sensory cues (Biswas et al. 2014; Knoferle et al. 2012; Krishna
30
2012), the present research is the first to focus on the effects of ambient light on food choices.
Moreover, while there has been extant work examining the effects of other sensory cues, such as
ambient scent and music, hardly any work has examined the effects of ambient light on product
choices. In fact, the findings of our research have implications for cross-modal influences,
whereby ambient light, which is processed through the visual system influences food choices,
similar predictions as those related to mental alertness. The findings of our research provide
process evidence in favor of the dominant role of mental alertness. The results of our studies did
not provide evidence for the role of inhibition or self-presentation focus as alternative underlying
processes. However, we just had self-reported measures of self-presentation focus and inhibition,
and in the case of inhibition, Study 1c had a response elicitation method that enhances inhibition
and yet the pattern of results from this study were similar to that of the other studies. In other
words, we did not empirically rule out the roles of inhibition and self-presentation focus as
underlying processes – we just found direct empirical evidence for the role of mental alertness.
In fact, it is possible that although mental alertness is the dominant underlying process, inhibition
and self-presentation focus might play supporting roles. This is especially likely since these
variables are interlinked (Baumeister et al.1994; Carver and Scheier 1978; Heatherton, Striepe,
and Wittenberg 1998; Heinz et al. 2011). Future research should examine the link between these
variables and their respective roles, in the context of ambient light luminance, in greater depth.
Along these lines, bright ambient light can potentially heighten social desirability concerns,
31
which in turn might influence food ordering. Additional research is needed to examine this in
greater depth.
While we did provide empirical evidence regarding the underlying process related to
mental alertness, one limitation of our studies (except Study 1b) is that alertness was measured
after participants made their food choices. This was done to avoid hypothesis guessing or
suspicion from responding to alertness questions before making the food choice. Accordingly,
the mediation effects demonstrated in our studies might be more correlational, than causal, in
nature. At the same time though, the objective measure of alertness, in Study 1b, was conducted
Our field study was conducted at a casual dining restaurant with food options that varied
significantly in terms of healthfulness levels and our lab studies were conducted with young
adults who have varied levels of health goals. The findings of our studies may not hold in the
context of fast food restaurants, where most of the consumers tend to visit with the mindset of
ordering something unhealthy; along similar lines, not surprisingly, frequency of fast-food
restaurant visits is associated with higher BMI levels (Rosenheck 2008). To further examine this
issue, we conducted an online (MTurk) study with thirty participants (57% females; average age
= 33 years), asking them whether they go to a fast-food place with the goal of ordering
something healthy or unhealthy. Overall, 90% (10%) of the participants indicated they go in with
a goal of ordering something unhealthy (healthy) (χ2 = 19.20, p < .01) when they visit fast-food
places. With such a strong tendency for ordering something unhealthy while visiting a fast-food
instinctively become more vigilant in dimly lit environments due to enhanced perception of
32
speculated that vigilance might be more at play in novel and/or unfamiliar environments. For
example, dining as a tourist at an unfamiliar restaurant in a new place can be a good catalyst for
enhancing vigilance level when the ambient light is dimmed. Can effects of enhanced vigilance
be consistent with those of enhanced mental alertness or can they offer competing hypotheses?
Future research needs to examine if the main effects of ambient lighting observed in our studies
might get changed for studies conducted in unfamiliar or novel settings. Similarly, self-
presentation focus might have been a more influential factor in dining contexts where impression
management is key. So for dining experiences in the contexts of dating or job interviews,
ambient lighting might influence self-presentation focus to a greater extent than that observed in
personality or behavioral traits, which can be relevant moderators. For instance, an individual’s
healthiness perspective can moderate the effects observed in our studies. Along similar lines, the
time of day might influence consumers’ alertness levels and also the luminance levels outside the
stores and restaurants. Moreover, time of day can be associated with different types of meals,
which in turn can influence consumption outcomes (Khare and Inman 2006). Hence, time of day
can potentially moderate the findings observed in our studies. Along similar lines, incidental
affect can potentially influence alertness levels and also food choices (Garg, Wansink, and
Inman 2007). Future research should examine the role of such potential moderators that can be
managerially relevant.
We examined choices involving healthy and unhealthy food items only. Future research
should examine the effects of ambient light luminance on non-food product choices in other
33
domains. For instance, would indulgences in terms of high priced purchases be greater when a
store has dim (vs. bright) ambient light? How about purchases of utilitarian versus hedonic
products across different ambient light luminance levels? Additional studies are needed to
We focused on the effects of ambient light. However, retail atmospherics entail several
other elements (such as music, scent, wall color, and overall décor). Hence, it might be
interesting to examine the interaction effects between ambient light and these other elements of
retail atmospherics. Along similar lines, while we study effects of ambient light luminance, there
are other elements of ambient light, such as color of the light, that are potentially interesting. For
instance, Lehrl et al. (2007) showed that keeping luminance levels constant, blue (vs. yellow)
light leads to greater alertness and faster information processing. Can this in turn influence
product choices? Given the dearth of research in the marketing literature examining the effects of
ambient light on product choices, this is relatively uncharted territory with potential for
Since managers of retail outlets and restaurants can usually control the retail atmospheric
elements (such as the luminance level of the ambient light) with relative ease, it is important to
understand how changing such elements influences choice. In essence, since ambient light
influences consumer choices, marketers can potentially manipulate the ambient light luminance
The findings of our research can be of interest to regulators as well. Regulators have
often expressed concerns about widespread obesity and have often tried to encourage healthy
34
eating through restrictive mechanisms, such as for example, in the recent case related to
restricted sales of sugary beverages in New York City (Strauss and Castagna 2013). However,
restrictive policies are often met with consumer reactance, as was the outcome in the New York
City beverage case (Saul 2013). Instead of having restrictive policies, which often lead to
consumer reactance, a more effective strategy to encourage healthy eating would be through the
use of subtle factors, such as ambient light. Since ambient light luminance influences choices
between healthy and unhealthy options, employing certain ambient light luminance levels can be
an effective strategy to influence choices for healthy versus unhealthy food items. In addition,
ambient light luminance being a subtle environmental factor and not being restrictive in nature,
While we demonstrate the effects of ambient light on food choices across a robust set of
experiments and thus provide existence proof of ambient lighting effects, care must be taken
when generalizing across all types of restaurant or store settings. As previously mentioned, there
can be several settings where the effects of our studies may not hold; for example, Dairy Queen
has very bright ambient lighting while some fine vegan restaurants have dim lighting. This is
especially likely to be the case since several other factors, apart from ambient light, such as
brand image, type of restaurant, and availability of options, among others, influence food
choices. Moreover, alertness levels can moderate the effects. That is, very high and very low
levels of alertness attenuate the effects, as demonstrated by Studies 2a and 2b and the study
In terms of consumer wellbeing, dining in brightly lit ambient settings might be a good
option if the goal is to enhance choice likelihood for healthy options. Since dim (vs. bright)
ambient light reduces mental alertness level, dining in dimly lit environments might lead to
35
greater likelihood of yielding to the temptation of going for the sumptuous, but unhealthy,
chocolate dessert. Hopefully, our study has shed some light on the effects of ambient light on
product choices and will trigger further research in this topic domain.
36
References
Amir, On and Jonathan Levav (2008), “Choice Construction versus Preference Construction:
(2), 145–58.
Areni, Charles S. and David Kim (1994), “The Influence of In-Store Lighting on Consumers’
Badia, P., B. Myers, M. Boecker, J. Culpepper, and J. R. Harsh (1991), “Bright Light Effects on
Body Temperature, Alertness, EEG, and Behavior,” Physiology & Behavior, 50 (3), 583–
88.
Baker, Julie, A. Parasuraman, Dhruv Grewal, and Glenn B. Voss (2002), “The Influence of
Multiple Store Environment Cues on Perceived Merchandise Value and Store Patronage
———, Todd F. Heatherton, and Dianne M. Tice (1994), Losing Control: How and
Biswas, Dipayan, Dhruv Grewal, and Anne Roggeveen (2010), “How the Order of Sampled
———, Lauren Labrecque, Donald Lehmann, and Ereni Markos (2014), “Making Choices
While Smelling, Tasting, and Listening: The Role of Sensory (Dis)similarity When
http://www.bosworthinstrument.com/Lightmeterinfo.html].
Broniarczyk, Susan M. and Wayne D. Hoyer (2006), “Retail Assortment: More ≠ Better,” in
Retailing in the 21st Century: Current and Future Trends, Manfred Krafft and Murali K.
Bruner, Gordon, C. (1990), “Music, Mood and Marketing,” Journal of Marketing, 54 (4), 94–
104.
Cajochen, Christian (2007), “Alerting Effects of Light,” Sleep Medicine Reviews, 11, 453–64.
Carroll, Matthew (2012), “How Retailers Can Replicate the ‘Magic’ of the Apple Store…
http://www.forbes.com/sites/matthewcarroll/2012/06/26/how-retailers-can-replicate-the-
magic-of-the-apple-store-online/].
Carver, Charles S. and Michael F. Scheier (1978), “Self-Focusing Effects of Dispositional Self-
Chandon, Pierre and Brian Wansink (2007), “The Biasing Health Halos of Fast Food Restaurant
Health Claims: Lower Calorie Estimates and Higher Side-Dish Consumption Intentions,”
——— and ——— (2012), “Does Food Marketing Need to Make Us Fat? A Review and
———, J. Wesley Hutchinson, Eric T. Bradlow, and Scott Young (2009), “Does In-Store
Marketing Work? Effects of the Number and Position of Shelf Facings on Brand
Attention and Evaluation at the Point of Purchase,” Journal of Marketing, 73 (6), 1–17.
38
Chernev, Alexander (2011), “The Dieter’s Paradox,” Journal of Consumer Psychology, 21 (2),
178-83.
Crowley, Stephanie, J., Clara Lee, Christine Y. Tseng, Louis F. Fogg, and Charmane I. Eastman
18 (6), 513–23.
Daurat, Agnes, Acacia Aguirre, Jean Foret, Philippe Gonnet, Anne Keromes, and Odile Benoit
(1993), “Bright Light Affects Alertness and Performance Rhythms during a 24-h
Dhar, Ravi and Klaus Wertenbroch (2012), “Self-Signaling and the Costs and Benefits of
Duke, Aaron A. and Laurent Bègue (2015), “The Drunk Utilitarian: Blood Alcohol
27.
Fenigstein, Allan, Michael F. Scheier, and Arnold H. Buss (1975), “Public and Private Self-
Fern, Edward F. (1982), “Why do Focus Groups Work: A Review and Integration of Small
Andrew Mitchell, Ann Arbor, MI: Association for Consumer Research, 444–51.
Gardner, Meryl P., Brian Wansink, Junyong Kim, and Se-Bum Park (2014), “Better Moods for
Better Eating? How Mood Influences Food Choice,” Journal of Consumer Psychology,
24, 320–35.
39
Garg, Nitika, Brian Wansink, and J. Jeffrey Inman (2007), “The Influence of Incidental Affect
Hayes, Andrew F. (2012), “PROCESS: A Versatile Computational Tool for Observed Variable
Heatherton, Todd F., Meg Striepe and Lauren Wittenberg (1998), “Emotional Distress and
Disinhibited Eating: The Role of Self,” Personality and Social Psychology Bulletin, 24
(3), 301–13.
Heinz, Adrienne J., Anne Beck, Andreas Meyer-Lindenberg, Philipp Sterzer, and Andreas Heinz
Herman, Peter C. and Janet Polivy (1993), “Mental Control of Eating: Excitatory and Inhibitory
Hirsch, Jacob B., Adam D. Galinsky, and Chen-Bo Zhong (2011), “Drunk, Powerful, and in the
Dark: How General Processes of Disinhibition Produce Both Prosocial and Antisocial
Irmak, Caglar, Lauren G. Block, and Gavan Fitzsimons (2005), “The Placebo Effect in
Janssen, Loes, Bob M. Fennis, Ad Th.H. Pruyn, and Kathleen Vohs (2008), “The Path of Least
Kasof, Joseph (2002), “Indoor Lighting Preferences and Bulimic Behavior: An Individual
Khan, Uzma, Ravi Dhar, and Klaus Wertenbroch (2005), “A Behavioral Decision Theory
Research on Consumer Motives, Goals and Desires, S. Ratneshwar and David Glen
Khare, Adwait and J. Jeffrey Inman (2006), “Habitual Behavior in American Eating Patterns:
Kivetz, Ran and Itamar Simonson (2002), “Self-Control for the Righteous: Toward a Theory of
Knoferle, Klemens, Eric R. Spangenberg, Andreas Hermann and Jan R. Landwehr (2012), “It is
All in The Mix: The Interactive Effect of Music Tempo and Mode on In-Store Sales,”
Krishna, Aradhna (2012), “An Integrative Review of Sensory Marketing: Engaging the Senses to
332–51.
Laran, Juliano (2010), “Goal Management in Sequential Choices: Consumer Choices for Others
(August), 304–14
41
Leary, Mark and Susan C. Atherton (1986), “Self-Efficacy, Social Anxiety, and Inhibition in
Kornhuber and S. Bleich (2007), “Blue Light Improves Cognitive Performance,” Journal
Lim, Julian and David Dinges (2008), “Sleep Deprivation and Vigilant Attention,” Annals of the
Lockley, Steven W., Erin E. Evans, Frank A.J.L. Scheer, George C. Brainard, Charles A.
Czeisler, Daniel Aeschbach (2006), “Short-Wavelength Sensitivity for the Direct Effects
Lowden, Arne, Torbjöjn Åkerstedt, and Roger Wibom (2004), “Suppression of Sleepiness and
Melatonin by Bright Light Exposure During Breaks in Night Work,” Journal of Sleep
Mattila, Anna S. and Jochen Wirtz (2001), “Congruency of Scent and Music as a Driver of In-
Meyers-Levy, Joan and Rui Zhu (2007), “The Influence of Ceiling Height: The Effect of Priming
on the Type of Processing That People Use,” Journal of Consumer Research, 34 (2),
174–86.
———, ———, and Lan Jiang (2010), “Context Effects from Bodily Sensations: Examining
Bodily Sensations Induced by Flooring and the Moderating Role of Product Viewing
Morrin, Maureen and S. Ratneshwar (2003), “Does It Make Sense to Use Scents to Enhance
Neuberg, Steven L., Douglas T. Kenrick, and Mark Schaller (2011), “Human Threat
Nielsen, Martin V., Svend Svendsen, Lotte B. Jensen (2011), “Quantifying the Potential of
Oberfeld, Daniel, Heiko Hecht, Ulrich Allendorf, and Florian Wickelmaier (2009), “Ambient
Lighting Modifies the Flavor of Wine,” Journal of Sensory Studies, 24, 797–32.
Pope, Lizzy, Andrew S. Hanks, David R. Just, and Brian Wansink (2014), “New Year’s Res-
Illusions: Food Shopping in the New Year Competes with Healthy Intentions,” PLOS
Preacher, Kristopher J. and Andrew F. Hayes (2008), “Asymptotic and Resampling Strategies for
Romero, Marisabel and Dipayan Biswas (2016), “Healthy Left, Unhealthy Right: Can
Displaying Healthy Items to the Left (versus Right) of Unhealthy Items Nudge Healthier
Rosenheck, Robert (2008), “Fast Food Consumption and Increased Calorie Intake: A Systematic
535–47.
43
Rüger, Melanie, Marijke C. M. Gordijn, Domien G. M. Beersma, Bonnie de Vries, Serge Daan
R1420.
Sala, Laurenne (2010), “Dining in the Dark at Opaque: Eat Like Nobody’s Watching,” (accessed
dark-at-opa_b_783383.html].
Salerno, Anthony, Juliano Laran and Chris Janiszewski (2014), “Hedonic Eating Goals and
Saul, Michael H. (2013), “Judge Cans Soda Ban,” Wall Street Journal, 03/11/2013,
[http://online.wsj.com/article/SB10001424127887323826704578354543929974394.html
].
Schaller Mark, Justin H. Park, and Annette Mueller (2003), “Fear of the Dark: Interactive Effects
of Beliefs about Danger and Ambient Darkness on Ethnic Stereotypes,” Personality and
Scheibehenne, Benjamin, Peter M. Todd, and Brian Wansink (2010), “Dining in the Dark: The
Importance of Visual Cues for Food Consumption and Satiety,” Appetite, 55 (3), 710–13.
Shiv, Baba and Alexander Fedorikhin (1999), “Heart and Mind in Conflict: The Interplay of
(3), 278–92.
44
———, Ziv Carmon and Dan Ariely (2005), “Placebo Effects of Marketing Actions: Consumers
May Get What They Pay For,” Journal of Marketing Research, 42 (November), 383–93.
Spangenberg, Eric R., Ayn E. Crowley, and Pamela W. Henderson (1996), “Improving the Store
Spence, Charles, Nancy M. Puccinelli, Dhruv Grewal, Anne L. Roggeveen (2014), “Store
88.
Steele, Claude M. and Robert A. Josephs (1990), “Alcohol Myopia: Its Prized and Dangerous
Strauss, Gary and Rebecca Castagna (2013), “NYC Ban on Large Sugary Drinks Fizzles Again,”
york-ban-on-larged-sized-sugary-drinks-rejected/2599429].
The Engineering Toolbox (2016), “Luminance - Recommended Light Levels,” (Last accessed:
d_708.html].
Thimijan Richard W. and Royal D. Heins (1983), “Photometric, Radiometric and Quantum Light
818–22.
Thomas, Maria, Helen Sing, Gregory Belenky, Henry Holcomb, Helen Mayberg, Robert
Dannals, Henry Wagner, David Thorne, Kathryn Popp, Laura Rowland, Amy Welsh,
Sharon Balwinski, and Daniel Redmond (2000), “Neural Basis of Alertness and
45
(4), 335–52.
Thorington, Luke (1985), “Spectral, Irradiance, and Temporal Aspects of Natural and Artificial
Verwey, Willem B. and David M. Zaidel (2000), “Predicting Drowsiness Accidents from
Personal Attributes, Eye Blinks and Ongoing Driving Behaviour,” Personality and
Virginie Gabel, Micheline Maire, Carolin F. Reichert, Sarah L. Chellappa, Christina Schmidt,
Vanja Homes, Christian Cajochen, Antoine U. Viola (2015), “Dawn Simulation Light
Wansink, Brian (2004), “Environmental Factors that Increase the Food Intake and Consumption
——— (2006), Mindless Eating Why We Eat More Than We Think. New York: Bantam Dell.
———, David R. Just, and Collin R. Payne (2009), “Mindless Eating and Healthy Heuristics for
———, Mitsuru Shimizu, Armand V. Cardello, and Alan O. Wright (2012), “Dining in the
Dark: How Uncertainty Influences Food Acceptance in the Absence of Light,” Food
——— and Koert van Ittersum (2012), “Fast Food Restaurant Lighting and Music can Reduce
Calorie Intake and Increase Satisfaction,” Psychological Reports: Human Resources &
Wardle, Jane, Anne M. Haase, Andrew Steptoe, Maream Nillapun, Kiriboon Jonwutiwes, and
France Bellisie (2004), “Gender Differences in Food Choice: The Contribution of Health
Weitz, Barton and Mary Beth Whitfield (2006), “Trends in U.S. Retailing,” in Retailing in the
21st Century: Current and Future Trends, Manfred Krafft and Murali K. Mantrala, eds.
White, Katherine, Lily Lin, Darren Dahl, and Robin J. B. Ritchie (2016), “When do Consumers
Wilcox, Keith, Beth Vallen, Lauren Block, and Gavan J. Fitzsimons (2009), “Vicarious Goal
Zhong, Chen-Bo, Vanessa L. Bohns, and Francesca Gino (2010), “Good Lamps Are the Best
Figure 1
Study 1a: The Effects of Ambient Light Luminance on Ordering of Healthy (vs. Unhealthy)
70% 65.38%
60%
PERCENTAGE OF PARTICIPANTS
52.44%
50% 47.56%
40% 34.62%
30%
20%
10%
0%
BRIGHT DIM
AMBIENT LIGHT LUMINANCE
48
WEB APPENDIX
DIPAYAN BISWAS
COURTNEY SZOCS
BRIAN WANSINK
ROGER CHACKO
1b Participants were given the option to choose between two food options (100-
calorie Oreos and chocolate covered Oreos). Participants were asked:
“If given a choice, which one of these food options would you prefer to have
right now?
__________ 100 Calorie Oreos
__________ Chocolate Covered Oreos”
going up to seven numbers in the third and fourth sets. The four sets of
numbers displayed were:
4 22 35 41 60
12 29 43 64 77 86
3 17 29 47 53 75 88
2 21 34 49 57 72 83
1c Participants chose between a chocolate bar and a granola bar. Samples of each
of these two items were displayed on a large table. Participants were asked:
“There are two food options in front of you – a piece of chocolate and a piece
of granola bar. If given a choice, which one of these food options would you
prefer to have right now?
_______ Chocolate
_______ Granola Bar
Please raise your hand and indicate your preference to the researcher so that
s/he can give you your preferred food option.”
2a Choice option between baked potato and fries. Participants were asked:
“If given a choice, which one of these food options would you prefer to have
right now?
__________ Baked Potato
__________ Fries”
2b Participants were given the option to choose raisins or M&Ms. They were
asked:
“If given a choice, which one of these food options would you prefer to have
right now?
__________ Raisins
__________ M&M’s”
Mental alertness was measured by asking participants: “How alert do you feel
right now?” (1 = not at all alert, 7 = very alert).
In the choice task, participants chose a food option from a menu. The menu is
provided in Web Appendix C below.
Also, across all studies, participants were asked to indicate their gender and age.
51
This study examined the effects of diminished alertness level. As mentioned in the main
body of our paper, prior literature categorizes alertness and sleepiness as being at conceptually
opposite ends of a continuum (Åkerstedt and Gillberg 1990; Lockley et al. 2006). That is, dim
(vs. bright) light enhances sleepiness and reduces alertness. In fact, multiple studies have
examined alertness through the lens of sleepiness (Kayumov et al. 2000; Thomas et al. 2000).
Accordingly, we examine if inducing sleepiness might alter the effects observed in Studies 1a-
1c. Specifically, we hypothesize that under regular mental alertness levels, when given a choice
between healthy and unhealthy options, consumers will have greater preference for healthy
options when ambient light luminance is bright (vs. dim); under diminished mental alertness,
these effects will get attenuated, whereby the preference pattern will be the same across bright
It might be noted though that although the same pattern of effects are being predicted by
H2 (in the main article text), the direction of the means would be different. That is, for H2, the
means for the “alertness enhanced” bright and dim conditions approached the means for “regular
alertness” bright condition, while we are predicting that the means for the “diminished alertness”
bright and dim conditions should approach the means for “regular alertness” dim condition.
Method
A 2 (ambient light luminance: bright vs. dim) X 2 (mental alertness level: regular vs.
alertness. While in Study 2b (in the main article text), we had mental alertness level enhanced by
asking participants to be more alert while taking the survey, such an approach would be
52
unrealistic when diminishing alertness. That is, asking participants to be less alert or to be sleepy,
while taking the survey, would arouse suspicion. Accordingly, we manipulated diminished
alertness level through a priming task. Specifically, at the beginning of the study, participants
were asked to try to think of a situation when they were extremely sleepy. Then, similar to the
approach used in prior studies (e.g., Chiao et al. 2006), participants were asked to write a short
essay describing the situation when they were extremely sleepy. They were also asked to
highlight why they were so sleepy. In the “regular” mental alertness condition, participants were
After this, participants were given a menu (see Web Appendix C for the items listed in
the menu). The menu had eight items, four of which were healthy (i.e., Veggies & Cucumber-
Dill Yogurt Dip, Green Salad, Grilled Chicken Wrap, Grilled Fish Wrap) and the other four were
unhealthy (i.e., Mozzarella Sticks, Loaded Fries, Fried Chicken Sandwich, Battered Fried Fish
Sandwich). All menu items were priced equally. Participants were asked which one item they
would like to order from the menu. So, while our other lab experiments (Studies 1b-2b in the
paper) were conducted in the context of making a choice from a set of two options, this study
involved choosing from a menu with a wider set of options. After indicating their menu choice,
participants indicated how sleepy they felt at that moment (1 = not at all sleepy, 7 = very sleepy)
and how mentally alert they felt at that moment (1 = not at all alert, 7 = very alert).
One hundred eighty eight students (average age 22 years; 50.8% females) from a major
US university participated in this experiment for course credit. Four participants either did not
make a choice from the menu or selected multiple items (implying they did not pay attention to
the instructions for the menu selection task) and these participants were not considered for the
choice or calorie analyses. However, all participants responded to all the other measures. The
53
procedure, with the exception of the alertness manipulation and food choices, was similar to
Main tests. Employing the more conservative backward logistic regression for the 2
(ambient light luminance) X 2 (induced mental alertness level) model revealed a marginally
significant interaction effect on food choice (Wald χ2 = 2.98, p = .08). Including gender as a
covariate keeps the effect unchanged (Wald χ2 = 2.92, p < .09). Using forward logistic regression
reveals a stronger 2 X 2 interaction effect on food choice (Wald χ2 = 6.10, p < .05). Again,
including gender as a covariate keeps the pattern of effects unchanged (Wald χ2 = 5.52, p < .05).
Overall, gender did not have a significant main effect on food choices (χ2 = 1.25, p = .26) and
there was no interaction effect between gender and ambient lighting on choice (p > .40).
Follow-up tests showed that choice of the healthy item was higher with bright (vs. dim)
ambient light when alertness level was regular (Pbright-light = 70.97% vs. Pdim-light = 45.76%; χ2 =
7.92, p < .01), but the effects got attenuated when mental alertness was diminished (Pbright-light =
42.42% vs. Pdim-light = 43.33%; χ2 = .01, p = .94). Consistent with our conceptualization, the
choice likelihood for healthy options was similar for “regular alertness” dim light condition and
the “diminished alertness” dim and bright light conditions (all p’s > .75). That is, diminishing the
alertness level brought choice patterns similar to the dim lighting condition.
We also examined the effects on total calories chosen. The results of a 2 (ambient light)
X 2 (induced alertness) ANOVA on total calories of the chosen option showed a marginally
significant interaction effect (F(1, 180) = 3.13, p < .08). Follow-up tests showed that in the
absence of any priming, total calories chosen was lower when ambient light was bright (vs. dim)
54
(Mbright-light = 323.19, SD = 178.37 vs. Mdim-light = 432.60, SD = 196.66; F(1, 180) = 9.40, p <
.01). In contrast, when mental alertness was reduced (through sleepiness priming), the total
calories chosen were same for bright versus dim ambient lighting (Mbright-light = 443.18, SD =
204.28 vs. Mdim-light = 444.57, SD = 220.70; F(1, 180) = .001, p = .98). These results support our
sleepiness (Mregular alertness = 4.17 vs. Mdiminished alertness = 3.78; F(1, 186) = 3.21, p < .08). Also, as
observed in our other studies, self-reported mental alertness level was lower when ambient light
was dim (vs. bright) (Mdim-light = 4.65 vs. Mbright-light = 5.11; F(1, 186) = 6.71, p < .05).
The results of this study again emphasize the role of mental alertness as the dominant
underlying factor in influencing the effects of dim versus bright ambient lighting on food
choices. While the paper’s Study 2a influenced mental alertness through a placebo effect and
Study 2b examined effects of enhanced mental alertness, the current study demonstrated the
Mozzeralla Sticks
$8.50
Fried mozzarella sticks served with marinara sauce.
Loaded Fries
Fries topped with cheddar and Colby jack cheeses, $8.50
sour cream and chives served with Ranch dressing.
Green Salad
Mixed greens, tomatoes, red and green peppers, $8.50
onions, and cucumbers with vinaigrette dressing.
Åkerstedt, Torbjöjn and Mats Gillberg (1990), “Subjective and Objective Sleepiness in the
Chiao, Joan Y., Hannah E. Heck, Ken Nakayama, and Nalini Ambady (2006), “Priming Race in
Biracial Observers affects Visual Search for Black and White Faces,” Psychological
Irmak, Caglar, Lauren G. Block, and Gavan Fitzsimons (2005), “The Placebo Effect in
Kayumov, Leonid, Vadim Rotenberg, Kenneth Buttoo, Christine Auch, S. R. Pandi-Perumal and
12 (1), 86-90.
Lockley, Steven W., Erin E. Evans, Frank A.J.L. Scheer, George C. Brainard, Charles A.
Czeisler, Daniel Aeschbach (2006), “Short-Wavelength Sensitivity for the Direct Effects
Thomas, Maria, Helen Sing, Gregory Belenky, Henry Holcomb, Helen Mayberg, Robert
Dannals, Henry Wagner, David Thorne, Kathryn Popp, Laura Rowland, Amy Welsh,
Sharon Balwinski, and Daniel Redmond (2000), “Neural Basis of Alertness and
(4), 335-52.