Available online at www.sciencedirect.

com

ScienceDirect
Journal of Consumer Psychology 24, 2 (2014) 217 – 225

Research Article
Why the bride wears white: Grounding gender with brightness☆
Gün R. Semin a,b,⁎, Tomás A. Palma c
a
Utrecht University, The Netherlands
b
Koç University,Turkey
c
CIS-ISCTE, Lisbon University Institute, Portugal

Received 6 March 2013; received in revised form 3 August 2013; accepted 7 September 2013
Available online 21 September 2013

Abstract

Two studies examine the grounding of gender by the alignment of the female–male with the bipolar dimension of light–dark (most likely due to
sexual dimorphism in skin pigmentation). We hypothesized and showed that in a speeded classification task male names are processed faster when
they are presented in a black typeface (Exp. 1) or a dark color (Exp. 2) than when they are presented in white or a light color, with the opposite
pattern for female names. The applied relevance of these findings is investigated in study 3 where lightness and darkness of consumables are
revealed to drive gender specific preferences for foods and drinks, with the lighter consumables being female and darker ones being male
preferences. Study 4 shows that gender preferences for consumer goods are uniformly driven by whether the good is in black or white, the former
being male and the latter being female preference. The implications of these findings are discussed for theory formation in relation to the grounding
of abstract concepts and in terms of how to design targeted marketing of products.
© 2013 Society for Consumer Psychology. Published by Elsevier Inc. All rights reserved.

Keywords: Grounding abstract concepts; Gender preferences; Metaphors; Embodied cognition; Situated cognition

Introduction within six months, after marketing and packaging were

Imagine a waiter bringing a cappuccino and an espresso
dutifully to a couple sitting at a table. Without being told who
ordered what, she serves the woman with the cappuccino and
the man with the espresso. The waiter will probably reach this
conclusion and execute it without giving it much thought. But
this example is only the tip of the iceberg. Take another
example. Coca Cola Light came in white cans and bottles. It
simply did not sale well among the coveted young male
population. Borrowing an idea from their Australian division
they introduced Coke Zero in black1 . The brand took off
☆ We would like to express our appreciation to the reviewers who gave most
helpful advice to revise, in particular, the theoretical section of this paper. We
would also like to thank Leonel Garica Marques for his statistical advice.
⁎ Corresponding author at: Utrecht University, Heidelberglaan 1, 3584 CS
Utrecht, The Netherlands.
E-mail address: g.r.semin@uu.nl (G.R. Semin).
1
See: www.ajc.com/news/business/coke-zero-becomes-a-hero-for-coca-cola-
co/nQkHh/).
1057-7408/$ -see front matter © 2013 Society for Consumer Psychology. Published by Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcps.2013.09.003
completed.
As we shall argue and show, there is a pervasive association
between gender and brightness, with bright marking female
and dark marking male, but for reasons other than a metaphoric
association. The aim of this research is not merely to reveal this
special association between brightness and gender, along with
the implications this association invites. It is also to show that
different contexts in which sensory-structurings are activated
determine not only the meaning that sensory grounding
acquires but also necessitates different conceptual explana-
tions of the grounding process, relying on the same sensory
dimension. One meaning of the sensory opposition between
brightness and darkness can be found in the well-documented
context of evil and good, or positive and negative, as used in
popular culture (e.g., Star Wars), remonstrated in religion
(e.g., Bible, Buddhist writings, Upanishad), and examined
in psychological research (e.g., Meier, Robinson, & Clore,
2004; Meier, Hauser, Robinson, Friesen, & Schjeldahl, 2007;
Sherman & Clore, 2009).
218 G.R. Semin, T.A. Palma / Journal of Consumer Psychology 24, 2 (2014) 217–225
However, as we shall argue and show, in the context of
gender, the association with bright and dark is orthogonal to
valence, or affect. Moreover, we shall argue that gender is not
metaphorically associated with the dimension of light–dark
but that there are ‘natural’ reasons for the association. Thus,
light vs. dark as a polar opposition functions as a sensory
structuring device to ground valence in one context (good vs.
bad). In another context that is orthogonal to valence, it
functions as a sensory structuring device to ground gender
(male vs. female) and is probably based on evolutionarily adaptive
and universal differences in skin color between males and females
(cf. Jablonski, 2004; Jablonski & Chaplin, 2000, 2002).
We start with a brief overview of the valence context where
the implications of the polarity shared by the valence and
brightness dimensions have been examined (e.g., Banerjee,
Chatterjee, & Sinha, 2012; Meier et al., 2004; Meier et al.,
2007; Sherman & Clore, 2009). Then, we turn to the current
focus, namely the association between gender and brightness
arguing that it is likely to derive from sexual dimorphism in skin
brightness. Subsequently, we furnish an overview of the research
consisting of 4 experiments that investigate not only the sensory
grounding of gender, namely bright with female and dark with
male (Exp.'s 1–2), but also the marketing implications (Exp.'s
3–4), namely how they influence decisions across a range of
choice dilemmas.
Background: light–dark evil–good negative–positive
Departing from Lakoff and Johnson's conceptual metaphor
theory (Lakoff & Johnson, 1980, 1999; also see Gibbs, 2006),
research on brightness and valence has been derived from the
contention that valence is possible as an abstract thought
because such a category (good–bad, positive–negative) relies
on our capacity for a sensory-based metaphor. The function of
the metaphor is to link the abstract category to physical
experience and thus structure human thought and shape the
way we encode, store, and retrieve information. Thus, by being
activated, metaphors provide sensory scaffolds for our thinking
about abstract domains like time (e.g., Boroditsky, 2000), justice
(e.g., Thibodeau & Boroditsky, 2011), morality (e.g., Lee &
Schwarz, 2011; Zhong & Liljenquist, 2006), and in this particular
case, affect or valence (e.g., Meier et al., 2004; Meier et al., 2007;
Sherman & Clore, 2009).
In this view, cognition and abstract categories such as
valence rather than being amodal are inherently body based
(e.g., Barsalou, 1999; Gibbs, 2006; Niedenthal, Barsalou,
Winkielman, Krauth-Gruber, & Ric, 2005; Smith & Semin,
2004; Semin & Smith, 2002, 2013; Semin, Garrido, & Farias,
in press; Semin, Garrido, & Palma, 2012, 2013; Wilson, 2002).
This take on abstract categories has led to research examining
the relationship between brightness and valence. For instance,
Meier et al. (2004) have reported that negative words were
evaluated faster and more accurately when presented in a black
font rather than a white font, and positive words were
evaluated faster and more accurately when presented in a
white rather than a black font. They suggest that affective
evaluations activate perceptions of stimulus brightness. Meier,
Robinson, Crawford, and Ahlvers (2007) report a series of studies
where, inter alia, participants had to match the brightness of a
word to a set of squares varying in shading. The results revealed a
metaphor-consistent bias in their brightness judgments. Thus,
much like space that grounds time (e.g., Boroditsky, 2000),
political concepts (e.g., Farias, Garrido, & Semin, 2013), valence
(cf. Crawford, 2009), or power (e.g., Schubert, 2005) the
brightness dimension can be deployed to ground different abstract
concepts, and one of these is gender. In short, sensory-structuring
is a situated process (Semin & Smith, 2013; Smith & Semin, 2004;
Semin, Garrido, & Palma, 2013; Semin et al., 2012, in press). In
one context light vs. dark can ground valence or good and evil. In
another context, this sensory dimension can ground gender. As we
shall argue, the way that the light–dark dimension grounds
gender is not necessarily based on any obvious metaphorical
association but a universal dimorphism in skin color between
males and females.
Beyond good and evil and towards male and female
There are a number of cultural traditions that associate
female and male and white and black or brightness. White is
the color most commonly associated with innocence and
virginity. The white virgin is the epitome of purity, grace, and
chastity. Black in contrast is culturally often seen as the color
of authority and seriousness and notably the symbols of black
that have to do with authority and seriousness are predomi-
nantly associated with the male gender. These and other related
cultural associations with maleness and femaleness are actually
likely to be based on the sexual dimorphism in skin pigmentation
as a result of natural selection (Jablonski & Chaplin, 2000). As
Jablonski (2004, p. 609) argues, the lighter skin pigmentation of
females facilitates the maximization of “…cutaneous vitamin D3
production in order to meet the absolutely higher calcium
requirements of pregnancy and lactation” (p. 609). Moreover, she
argues that “… darker pigmentation may have been the object of
natural selection in males because of the importance of
maintaining optimal levels of folate in order to safeguard sperm
production, a process depending on folate for DNA synthesis”
(p.609). Thus, the different accounts that have been advanced for
the emergence of this sexual dimorphism such as females being
more attractive if they have a lighter pigmentation (Frost, 1988)
or the diverse cultural representations about lighter pigmentation
or brightness being indicative of virginity, purity and so on may
have contributed to the sexual selection process, but are unlikely
to be their ultimate cause (cf. Jablonski & Chaplin, 2000). In
essence, “human skin color has evolved to be dark enough to
prevent sunlight from destroying the nutrient folate but light
enough to foster the production of vitamin D” (Jablonski &
Chaplin, 2002, p. 76). The significance of the sexual dimorphism
in skin pigmentation is complemented by studies on gender
recognition from faces. These suggest that an important feature in
the gender recognition process is color, with male faces being
darker than female faces (e.g., Nestor & Tarr, 2008; Nestor,
Vettel, & Tarr, 2008).
These considerations invite a theoretical grounding of the
gender-brightness link other than a metaphor driven one. Indeed,
 G.R. Semin, T.A. Palma / Journal of Consumer Psychology 24, 2 (2014) 217–225
this evolutionary account finds a better theoretical home in
perceptual symbol systems (Barsalou, 1999, 2008), according to
which simulations of previous experiences are central to
cognition. Thus, repeated experiences of lighter skin pigmenta-
tion for females and darker pigmentation of males may have
become the basis of a gender based dimorphic perceptual
representation, which could have resulted in an unconscious
neural representation of this visual input. These in turn could
have been transduced into varying conscious experiences or
representations shaped by diverse cultural practices that could
have emerged, as for example in the case of brides wearing white,
the color of purity and virginity and grooms wearing a dark, often
black outfit, inter alia.
In the research we report, the first two studies investigate
whether this sexual dimorphism in skin pigmentation is
reflected in gender representations, by examining whether
gradations in shading differentially facilitate the speed of
gender classifications. Obviously, these studies do not test any
links between systematic skin pigmentation differences be-
tween males and females and their explicit representation.
Nevertheless, they examine the association between gender and
their sensory grounding on the light–dark dimension.
Overview
In the research we report and present two groups of
experiments. The experiments in the first group were designed
to examine the sensory grounding of gender. The second group
consisted of experiments extending this research to an applied
setting.
The first group consisted of 2 experiments that examined
whether presenting male and female names in light and dark
facilitates their classification. In the first experiment, male and
female names that had been piloted for valence were presented
in either black or white against a gray background and
participants' task was to classify whether the name was a
male or a female name. In the second experiment, the same
paradigm is used however we presented the names in 4 shades
of green. The hypothesis under examination was that the
classification of male names relative to female names is
facilitated when they are with a dark typeface and the reverse
was predicted for female names. When female names are
presented in light typeface then they are classified faster than
male names.
The first two experiments were designed to define the sensory
structuring of gender. The systematicity that is unfolded in these
studies is informative about how gender is grounded. The
question that one can raise is: what use can this novel insight be
put to? This is the task of the second group of studies, which
consisted of two experiments. The first one was about choice
behavior between different food items (e.g., espresso vs.
cappuccino) that differ in terms of their light and dark features.
The participants' task was to decide ‘whom to give what’
between a male and a female character. The second experiment
has a shopping scenario where a man and a woman were to buy
different objects (e.g. coffee machine, scooter) and the partici-
pants' task was to decide which of the two objects was more
219
likely to be bought by the male character and which by the female
character. The objects in each choice situation were either light or
dark.
Experiments 1–2: dark male–light female
The first two experiments examined the sensory grounding
of gender hypothesis. The common denominator to these
experiments is the experimental paradigm, which is a simple
classification task with a response window of 800 ms within
which participants have to decide and indicate whether a name
presented on the monitor is a male or a female name. The
specific hypothesis under examination posits that male names
presented in darker typeface will be classified faster than the
same names in lighter typeface. For female names, precisely the
opposite is predicted. Female names presented in lighter
typeface will be classified faster than those in darker typeface.
Experiment 1: male is black and female is white
In the first experiment, male and female names that had been
matched for valence via a pilot study were presented to
participants in either black or white typeface. Since the task was
an easy one, we gave participants a response window of 800 ms.
Responses were given by means of a response box.
Method
Participants and stimulus material
Thirty-five students (23 women, mean age = 22.2) participated
as paid volunteers (€2). The experimental stimuli consisted of a
set of 40 names half of which were typical Dutch male names and
the other half female names (Appendix A). In a pilot study with 22
participants (who did not participate in the main experiment), we
compared the valence of the 20 male and 20 female names,
whereby each name had to be judged on a 7-point scale, the
anchors of which were negative (1) and positive (7). The
evaluation of the male names (M = 4.78, SD .54) did not differ
from the evaluation of the female names (M = 4.97, SD = .57)
t(21) = − 1.44, p = .17.
Procedure
Upon arrival, participants were led to individual cubicles
where the experiment was presented as a computer adminis-
tered task. The task of the participants was simple: to classify
whether the name they saw on the monitor was a male or a
female name. The response keys were counterbalanced across
participants: half of the participants classified the male names
using the right key and the female names using the left key
while for the other half of the participants the key-labels were
reversed. The experimental session was preceded by a set of 8
practice trials with 2 male and 2 female names that did not appear
in the main experiment. First, participants saw a fixation point in
the middle of the monitor against a gray background. Then a
name appeared for 800 ms — namely the time window within
220 G.R. Semin, T.A. Palma / Journal of Consumer Psychology 24, 2 (2014) 217–225

which a response had to be given. The dependent variable was the

mean reaction time (RT) required to classify the name on the
monitor as male or female. After each response there was a gray
screen for 500 ms. Participants were instructed to be as fast and
accurate as possible. After each name, which was either in black
or white letters, participants received feedback when they were
slow (‘too slow’). Participants received feedback only during the
practice trials. Then participants received the actual set of trials.
The presentation was identical to the practice trials. They
received 80 trials where half of the names were presented in
white and the other half in black in a random order varying across
participants. Participants were thanked and debriefed after the
experiment was over.
Fig. 1. Mean reaction times in milliseconds as a function of color and gender of
name. Note. The error bars indicate 95% confidence intervals.
Results

The data were analyzed in a 2 × 2 repeated measures ANOVA Method

with names (male, female) and letter color (black, white) as the
two variables. The reaction times (RTs) for incorrect responses Participants
(include wrong responses and misses, namely participants who
could not respond within the 800 ms response window: 4.86%) Thirty-seven students (28 women, mean age = 22.19)
and the reaction times that deviated more than 3 standard participated as paid volunteers (€2).
deviations from the participants' condition mean (3.01% of the
correct RTs) were excluded from all analyses2. The predicted
Design and results
interaction between names and letter color was significant, F(1,
34) = 13,11, p b .001, ηp2 = .29. Fig. 1 reveals this pattern,
The experimental paradigm was identical in every respect,
showing that male names in black (M = 469 ms, SD = 33) were
except that the names were presented now in 4 shades of green
classified significantly faster than male names in white, (M =
in two blocks of 80 trials each. In each block each male and
479 ms, SD = 32), t(34) = 3.73, p b .001. The reverse is
female name was presented two times in two different color
observed for female names, with white female names being
shades. We selected 4 greens from a RGB color palette with the
classified marginally faster (M = 473 ms, SD = 41) than black
codes: light green: 0, 255, 0; medium light: 0, 191, 0; medium
female names (M = 482 ms, SD = 42), t(34) = 1.78, p = .08.
dark: 0, 128, 0; dark green: 0, 64, 0.
These results were not moderated by participants' gender,
Thus, in total each name was seen four times, one in each
F(1, 33) = .03, p = .85, nor by counterbalancing of the
shade of green. The experiment consisted of a name (male vs.
response key, F(1, 33) = 1.17, p = .29.
female) by letter color shades (light vs. medium light vs.
medium dark vs. dark) design. One participant appeared to have
reversed the response keys as he had only 1% of correct
Discussion and experiment 2: gender is a shady business
responses. Another two participants had accuracy scores below
62%, which was three times less than the standard deviation
The results of the first experiment provide evidence in favor
from the mean response accuracy (87%). Data from these 3
of the hypothesis. Male names in black and female names in
participants were removed from all analyses. The data were
white are processed significantly faster than when male names
analyzed in a 2 × 4 repeated measures ANOVA, with names
are presented in white typeface and female names in black
(male, female) and letter color shades (light, medium light,
typeface. If however the general argument relies on brightness
medium dark, dark) as the two variables.
vs. darkness, then what should matter in the classification of
We excluded reaction times (RTs) for incorrect responses
masculine and feminine names is the shades of lightness or
(missing values and responses not within the time window:
darkness in which the names are presented. Thus, in the second
11.41% of all RTs) and reaction times that deviated more than 3
experiment we decided to choose a particular color which—as
standard deviations from the participants' condition mean
far as we could ascertain—did not have any associations with
(1.95% of the correct RTs)3. The mean reaction times are
gender: green. We used the very same experimental paradigm
presented in Table 1. The pattern of results was similar in both
as in the first experiment, but now used four different shades of
blocks and thus the data of these blocks was analyzed together,
green instead of black and white.

2 3
Using the criterion of 2.5 standard deviations to remove outliers does not Using the criterion of 2.5 SDs to remove outliers, F(3, 99) = 11.39,
change the pattern of results, F(1, 34) = 12.44, p b .001, η2p = .27, neither it p b .001, η2p = .26, or log transforming the data, F(3, 99) = 9.00, p b .001,
does log transforming the data, F(1, 34) = 8.90, p b .001, η2p = .21. η2p = .21, does not change the results.
 G.R. Semin, T.A. Palma / Journal of Consumer Psychology 24, 2 (2014) 217–225
F(3, 99) = 1.48, p = .22. There was a significant main effect for
the color shades, F(3, 99) = 2.68, p = .05, ηp2 = .07. A Post hoc
test (i.e., Sheffé) showed no reliable differences between the
four-color shades (all p's N .16). As predicted, the interaction
was significant, F(3, 99) = 6.01, p b .001, ηp2 = .15. According
to our hypothesis, one should expect to obtain three significant
results: a negative linear trend between the classification of the
male names and the four intensities of green; a positive linear
trend between the classification of the female names; and,
importantly, a difference between these two linear trends. Thus,
not only should participants be faster classifying male names in
dark green than light green and faster classifying female names in
light green than dark green, but they also should be faster for male
names than female names in dark green and faster for female
names than male names in light green.
We carried out a linear trend analysis first for the male
names that showed a significant effect, t(33) = 2.81, p b .001.
As predicted, the male names were classified faster when
presented in dark green (M = 453 ms) than when presented in
light green (M = 465 ms) (see Table 1). We also conducted a
liner trend analysis for female names, which showed a
marginally significant effect, t(33) = 1.94, p = .06. This
analysis revealed that female names were classified faster
when presented in light green (M = 453 ms) than when
presented in dark green (M = 460 ms). A difference contrast
between the two linear trends (weights: male names = 1,
female names = − 1; light green = 3, medium light = 1,
medium dark: − 1, dark green = − 3) revealed a significant
effect, t(33) = 3.79, p b .001, thus confirming our prediction4 .
These results were not moderated by participants' gender,
F(3, 99) = .55, p = .65, nor by the counterbalancing of the keys
of the response board, F(3, 99) = .81, p = .49.
Discussion
The second experiment confirms that brightness of the
typeface in which names are presented influences the speed
with which names are classified as male or female. The darker the
shade of green in which the typeface is presented, the faster male
names are classified and the slower female names are classified.
The reverse is the case for lighter typefaces. The lighter the
green-shade of the typeface, the faster the female names are
classified, and the slower the male names are classified. In both
cases, we find the relevant significant linear trends. Importantly,
4
We also tested for the quadratic contrasts, which were significant for male
names, t(33) = 2.41, p = .02, and female names, t(33) = 2.49, p = .02.
However, please note that these significant quadratic effects simply mean that,
for both male and female names, the mean reaction times to names presented in
the two mid green shades are not equally distant from the mean reaction times to
the names presented in the two most extreme green shades. If these quadratic
effects would contribute to provide an alternative explanation to our
hypotheses, then a significant difference between the male and female names
quadratic trends would necessarily emerge (weights: male names = 1, female
names = − 1; light green = 1, medium light = − 1, medium dark: − 1, dark
green = 1), which was not the case, t(33) = − 1.03, p = .31.
221
Table 1
Mean reaction times in milliseconds of names as a function of gender and shade
of green typeface.
Shades of green
Very light Light Dark Very dark
Male names 465 (43) 456 (40) 452 (48) 453 (48)
Female names 453 (46) 447 (47) 451 (46) 460 (47)
Note. Standard deviations are in parentheses.
we found a difference between these two linear trends showing
that names presented in dark green are classified faster when they
are male names rather than female names and the opposite is
found in the case of names presented in light green. Overall, the
current findings from the first two experiments confirm the
general hypothesis that light is associated with females and dark
with males.
These findings suggest that Coca Cola Light and Coca Cola
Zero are designed perfectly—light and dark—to target the
populations they want to reach. Consequently, the light–dark
dimension, when used systematically, should bias decisions
and inferences in a gender specific way. This is what we turned
to in the next two studies.
Marking gender with brightness: so, what does it buy?
The two studies we have reported so far show that light vs.
dark as a polar opposition functions as a sensory structuring
device to ground gender (male vs. female). Revealing the
association between light vs. dark and female vs. male is one
thing; another thing is to inquire about what relevance such
knowledge may have. In the introduction to this paper we gave
the hypothetical example of a waiter bringing a cappuccino and
an espresso dutifully to a couple sitting at a table and
automatically giving the cappuccino to the female customer and
the espresso to the male customer without being told to do so. The
question we pursued in the next two studies simply inquired
whether the brightness dimension would influence gender-based
preference decisions.
Experiment 3: gender and consumption preferences I
Is there a distinctive gender driven preference when confronted
with a choice between dark vs. light beverages or food? In this
study, we gave participants a set of choices such as an espresso
and a cappuccino and asked which would be the preference of a
man (or woman). The hypothesis we pursued was that participants
would be more likely to allocate the darker option in the choice
dilemmas to the male target and the lighter option to the female.
Method
Participants, stimulus material and procedure
Fifteen students (10 women, mean age = 22.93) participated
in this study as paid volunteers (€2). They were presented with
222 G.R. Semin, T.A. Palma / Journal of Consumer Psychology 24, 2 (2014) 217–225
13 different situations (8 target situations and 5 filler situations)
where a man and a woman were always confronted with two
choices. In the target situations, the eight pairs of choices were
light and dark beverages and food (e.g., cappuccino vs.
espresso, milk vs. chocolate-milk, white wine vs. red wine),
while in the filler situations the five pairs of choices were of
different colors (the 8 critical dilemmas are presented in the
Appendix B). These situations were presented one by one, in
the computer screen, in a random order. Participants' task was
to indicate, for each situation, who (man or woman) chose or
got the dark option (e.g., “A man and a woman are sitting at a
bar and order a glass of red wine and a glass of white wine but
do not tell the barman who is supposed to get what. Who do
you think the barman serves the red wine when he brings the
drinks?”). Participants pressed the key “m” of the keyboard for
man and the key “v” for woman (“vrouw” in Dutch). The task
had no time limit. They were then debriefed, paid, and thanked.
Results and discussion
In Table 2 we present only the proportion of ‘man’ responses
to each of the eight critical dilemmas (see Appendix B).
Because participants had to choose only one of two response
options, that means the proportion of woman responses is equal
to one minus the proportion of man responses. For each
proportion, we included its associated 95% confidence interval
(CI). If the lower bound of the CI's does not include the
proportion of .50 then one can conclude with 95% of
confidence that participants chose significantly more often the
option ‘man’. If the lower bound of the CI's includes .50 then
there are no significant differences between the proportions of
man and woman responses. Because in all our dilemmas the
question formulated was always about the darker option, our
prediction was that the proportion of ‘man’ responses would be
significantly different from the proportion of ‘woman’
responses. Our expectations were confirmed in five of the
eight dilemmas (i.e., dilemmas 1, 3, 4, 5 and 6). In the other
three dilemmas (i.e., dilemmas 2, 7, and 8), there were no
significant differences between ‘man’ and ‘woman’ responses.
Table 2
Proportion of man and woman responses to each of the eight critical dilemmas.
Dilemmas
Dil.1 Dil. 2 Dil. 3
Man 0.80 0.60 0.93
−/+ 95% CI 0.5191; 0.3229; 0.6805;
0.9567 0.8366 0.9983
Dil.5 Dil. 6 Dil. 7
Man 0.87 0.80 0.33
−/+ 95% CI 0.5954; 0.5191; 0.1182;
0.9834 0.9567 0.6162
Note. 95% confidence intervals are presented for each proportion.
To further test our hypothesis we performed a general test
across all dilemmas. We subtracted .50 from the proportion of
man responses to each of the eight critical dilemmas and
submitted these to a one-way ANOVA repeated measures. If
the participants attributed more often the darker food options to
the man then the proportion of man responses should be
different from zero, that is, the intercept of the ANOVA should
be significant, which was the case, F(1, 14) = 27.43, p b .001,
ηp2 = .66.
One possible criticism to this experiment is that the
beverages and the food that we used in our critical dilemmas
vary across several dimensions, like taste or shape, and the
light/dark dimension is simply another one of these dimen-
sions. For instance, one can argue that some of the dark
beverages presented (e.g., coffee) are stronger than some of the
light beverages (e.g., cappuccino) and that men are stereotyp-
ically seen as stronger and women as fragile. Thus, participants
might be basing their responses on this stereotypical knowl-
edge about men and women. Although we find this argument
difficult to apply to all the critical dilemmas, we ran a pilot
study with a different group of 24 participants (18 women,
mean age = 22.20) to whom we randomly presented all the
food items used in the dilemmas and asked them to rate them
on two 7-point scales: strength (1 — weak; 7 — strong) and
brightness (1 — light, 7 — dark). These scales were randomly
presented for each food item. The results showed that indeed the
darker food items (M = 5.69) were not only judged as being
darker than the light food items (M = 2.37), t(23) = 14.79,
p b .001, but also that the darker food items were judged as
stronger (M = 5.02) than the light food items (M = 4.01),
t(23) = 5.35, p b .001. However, when we compute difference
scores between the light–dark ratings (M = 3.32) and the weak–
strong ratings (M = 1.02) and compare them, we confirm that the
former are significantly different from the second, t(23) = 9.51,
p b .001. Thus, this result suggests that the brightness dimension
plays a more important role in distinguishing between food items
than does the strength dimension. However, it is possible that the
differences in strength nevertheless bias the results to a greater
extent than the differences in brightness. To rule out this possible
confound between brightness and strength and extend our
findings to other products rather than food items, we conducted
a further experiment with a different set of objects with their
brightness being the only feature varying between them.
Experiment 4: gender and consumption preferences II
In the previous experiment we showed that participants
Dil. 4
systematically inferred that the darker food item or beverage of
the pair would be the man's choice rather than the woman's.
0.80
However, in this experiment we used different items to form the
0.5191;
0.9567 pairs and thus it is possible that participants were using other
dimensions to base their responses rather than brightness. To rule
Dil. 8 out this potential confound, in our fourth and last experiment we
0.47 presented participants with pairs of objects that were exactly the
0.2127; same except for their color: for half of the object pairs one of the
0.7341 pairs was in black and the other in white, while the other sets of
pairs (fillers) were in different colors.
 G.R. Semin, T.A. Palma / Journal of Consumer Psychology 24, 2 (2014) 217–225 223

Method Table 3
Proportion of man and woman judgments for the black consumer items.
Participants, stimulus material and procedure Objects
Black Black Black Black
Thirty students (21 women, mean age = 22.43) participated in Watch Glasses iPhone iPod
this experiment as paid volunteers (€2). They were presented
Man 0.93 0.83 0.90 0.90
with a cover story about two friends, Mark and Sophie, who had −/+ 95% CI 0.7793; 0.6528; 0.7347; 0.7347;
similar tastes, and who went shopping together. On the monitor, 0.9918 0.9436 0.9789 0.9789
the participants saw two similar objects side-by-side. There were
in total 16 pairs of objects. Eight pairs were in black and white Black Black Black Black
(target objects) and 8 pairs were in different colors (fillers). The Macbook Nespresso Scooter Bag
participants' task was to choose which of the two objects was Man 0.97 1.00 0.93 0.83
more likely to be bought by ‘Mark’ and which was more likely to −/+ 95% CI 0.8278; 0.8843; 0.7793; 0.6528;
be bought “Sophie”. Below each object there were two options — 0.9992 1.0000 0.9918 0.9436
“Mark” or “Sophie”. The participants' task was to press the key Note. 95% confidence intervals are presented for each proportion.
corresponding the desired option. They were asked first to decide
about the object on the left of the screen and then about the object
on the right of the screen. They were explicitly told that if they
chose “Mark” first then they had to respond “Sophie” to the
second object. The pairs of objects were presented in a random the white-female and black-male associations as critical
order and stayed on the screen until the participant provided the features driving assumed consumer preferences.
two responses5. The critical objects that were in black and white
were: A watch, pair of glasses, iPhone, iPod, Macbook, Nespresso General discussion
machine, scooter, and a bag.
The research we reported was designed to address two
general issues. The first one was to contextualize sensory
Results structuring. Earlier research on valence revealed that brightness
is a structuring device (metaphoric association: source) for
As can be seen in Table 3, which represents the preference valence (abstract target). The current research shows that the
judgments of the black consumer items, participants consistently context in which sensory-structuring is activated, determines
and nearly unanimously attributed the black products to Mark the meaning that brightness structuring acquires. While the
rather than to Sophie. The pattern of results is similar across the earlier research was based on the concept of valence the current
eight critical objects. In none of these objects did the proportion research is in the concept of gender. Brightness activated in the
of Mark responses contain the value of .50 of the 95% confidence context of valence or affect structures positivity and negativity
interval. As in the previous experiment we subtracted .50 to each in bright and dark respectively as the evidence to date shows
proportion and submitted these to a one-way repeated measures (e.g., Meier et al., 2004; Meier et al., 2007; Sherman & Clore,
ANOVA, which revealed a highly significant intercept, F(1, 2009). However, as we have shown in the first two
14) = 312.68, p b .001, ηp2 = .91. Thus, these results unambig- experiments, in the context of gender bright and dark structures
uously underline the white-female and black-male associations as gender. Moreover, the target names we chose were controlled
critical features driving assumed consumer preferences. for valence. Thus, the function of brightness in the context of
gender is orthogonal to valence. Names in white typeface in
Exp. 1 and in bright green in Exp. 2 facilitated the classification
Discussion of female names relative to the classification of male names.
The classification of male names was facilitated when they
As can be seen in Table 3, which represents the preference were presented in black typeface (Exp. 1) or dark typeface
judgments for the black consumer items, the assumption was (Exp. 2). Indeed, the black and white effects on gender
overwhelmingly that the male target would choose these items. classification obtained in Experiment 1 were generalized with
The white versions of the consumer items mirror of course Experiment 2 in which we used graded shadings of green
these percentages. Thus, the results unambiguously underline varying from light to dark green.
In short, we were able to show that light vs. dark is a
structuring device to ground gender (male vs. female). This
alignment, namely between light–dark and female–male,
5
emerges most probably because it is anchored in the sexual
Due to a programming error the darker product was always presented to the dimorphism of skin pigmentation that was the result of
left of the screen and spatial position of color was not randomized as was
initially intended. The exception was the pair white iPhone–black iPhone where natural selection (cf. Jablonski & Chaplin, 2000, 2002). It is
the white iPhone was presented to the left of the screen. The results for this pair likely that this systematic difference between males and
showed exactly the same pattern as for the other critical pairs. females in skin pigmentation is also at the root of diverse
224 G.R. Semin, T.A. Palma / Journal of Consumer Psychology 24, 2 (2014) 217–225

cultural associations between the light and dark and gender. Appendix A. Valence ratings of the male and female names
Such cultural groundings can be seen in a diversity of cases used in Experiments 1 and 2
such as the tradition that the bride wears white and the groom
wears black; white being associated with virginity; the Names N Mean Min Max SD
association between white and purity found in numerous Arno* 22 3.45 1.00 7.00 1.30
civilizations. For instance, in Greek civilization white was Bert* 22 4.32 2.00 7.00 1.43
associated with mother's milk. Milk, the product of the Bram* 22 4.50 2.00 7.00 1.74
Daan* 22 5.18 1.00 7.00 1.76
female of species, was regarded as a sacred substance in the Frank* 22 5.05 2.00 7.00 1.36
Talmud and in ancient Egypt white was connected with the Jesse* 22 4.91 3.00 7.00 1.34
goddess Isis. In India white symbolizes purity, divinity, Johan* 22 4.82 2.00 7.00 1.44
detachment and serenity. Across different eras and cultures Lars* 22 4.09 1.00 7.00 1.93
the color white has been associated with purity, cleanliness Lucas* 22 5.00 3.00 7.00 1.51
Luuk* 22 4.77 2.00 7.00 1.66
and femininity.
Martijn* 22 5.45 1.00 7.00 1.74
Having said that, the question arises as to whether Max* 22 5.23 3.00 7.00 1.63
such situated structural alignment has some real Nick* 22 4.77 1.00 7.00 1.95
life consequences. This was the objective of the last two Pieter* 22 4.27 1.00 7.00 1.67
studies, namely examining the potential application of this Ruben* 22 5.18 1.00 7.00 1.65
Stijn* 22 4.45 1.00 7.00 1.95
sensory structuring of gender uncovered in the first two
Thijs* 22 4.64 1.00 7.00 1.81
experiments. Thomas* 22 5.50 1.00 7.00 1.68
The last two demonstrations (since they are strictly speaking Tim* 22 5.68 2.00 7.00 1.25
not experiments) were devised with a view to examine whether Wouter* 22 4.27 1.00 7.00 1.72
brightness influences gender related inferences in hypothetical Amber 22 4.68 2.00 7.00 1.55
Anna 22 5.55 2.00 7.00 1.30
choice and shopping situations. This is returning to an
Anneke 22 4.68 1.00 7.00 1.86
examination of the question of Coca Cola Light and Coca Anouk 22 4.68 1.00 7.00 1.78
Cola Zero we mentioned in the opening to this paper and the Britt 22 4.14 1.00 7.00 1.75
influence the respective advertisement campaigns had. Indeed, Chantal 22 3.77 1.00 7.00 2.18
the last two studies illustrate the power of the light–dark Emma 22 5.00 1.00 7.00 1.75
Eva 22 5.55 2.00 7.00 1.44
sensory anchoring of different objects, may these be consum-
Fleur 22 5.00 1.00 7.00 1.93
ables (Experiment 3) or consumer objects (Experiment 4). Jasmijn 22 5.50 3.00 7.00 1.30
Indeed, the light vs. dark consumables led participants to Julia 22 5.14 1.00 7.00 1.46
allocate food or drink choices to females or males respectively. Lieke 22 4.64 2.00 7.00 1.71
More impressively, the choice of consumer objects (Experi- Lisa 22 5.45 2.00 7.00 1.37
Lotte 22 5.36 1.00 7.00 1.62
ment 4) shows an overwhelming gender preference — white
Maud 22 4.05 1.00 7.00 2.10
objects are female choices and black ones are male choices. Sanne 22 5.36 2.00 7.00 1.50
Of course, the last two studies utilized impoverished Sara 22 5.50 1.00 7.00 1.74
situations where no further variables interfered and partici- Sophie 22 5.86 3.00 7.00 1.08
pants were forced to make a choice between two alternatives. Tess 22 4.14 1.00 7.00 1.70
Íris 22 5.45 3.00 7.00 1.14
In the context of this research, the aim was to maximize the
impact of the sensory dimension. In a real choice situation, the Note. Names marked with an asterisk are male names.
elements entering decisions are much richer and the influence
of the light–dark dimension may be less so. However, the Appendix B. Critical dilemmas
purpose of research is of course to see the effects of a
particular variable (here: light vs. dark; white vs. black) on 1- A man and a woman are sitting in a restaurant and order an
choices or inference. Having said that real life situations espresso and a cappuccino, without informing the waiter
involve contexts that are richer and nevertheless we find that who wants what. Who do you think the waiter served which
the implications of the light–dark sensory structuring of drink?
objects are an important and successful message for gender 2- A man and a woman are having breakfast. There is a glass of
targeting of consumer goods. Indeed, advertising companies milk and a glass of chocolate milk on the tables. Who do you
appear to be aware of this—devised implicitly or explicitly, think is having the chocolate milk?
Coca Cola advertisement in the UK showed Coca Cola Light 3- A man and a woman are preparing their toasts for breakfast.
with an attractive female dressed in white next to Wayne On one toast there is salami, and the other has slices of
Rooney, the famous Manchester United soccer player dressed turkey filet. Who do you think will be having the toast with
in black—each holding a bottle of Coca Cola, hers being Coca salami?
Cola Light and his Coca Cola Zero. The sales of Coca Cola 4- A man and a woman are sitting at a bar and order a
seem to have taken off within six months after the introduction Coca-Cola and a Sprite but do not tell the barman who is
of Coca Cola Zero in black. Light and dark seem to illuminate supposed to get what. When the barman brings the drinks:
the way to targeted marketing. who does he give which drink?
 G.R. Semin, T.A. Palma / Journal of Consumer Psychology 24, 2 (2014) 217–225
5- A man and a woman are in a restaurant and are ordering the
main course. The one main course is with sate sauce, and the
other one with a champignon source. They do not inform
the waiter who wants what. Who do you think the waiter
served the dish with the sate sauce when he brought the
dishes?
6- A man and a woman are sitting at a bar and order a glass
of red wine and a glass of white wine but do not tell the
barman who is supposed to get what. Who do you think
the barman serves the red wine when he brings the
drinks?
7- A man and a woman are in a restaurant and order pancakes.
One with sugar and the other with cinnamon but do not tell
the waiter who is supposed to get what. Who does the
waiter give the one with cinnamon when he brings the
pancakes?
8- A man and a woman are at a supermarket in front of the
bread stand. One of them takes the whole wheat bread and
the other the corn bread. Who do you think chooses the
whole wheat bread?
References
Banerjee, P., Chatterjee, P., & Sinha, J. (2012). Is It Light or Dark? Recalling
Moral Behavior Changes Perception of Brightness. Psychological Science,
23, 407–409.
Barsalou, L. W. (1999). Perceptual symbol systems. The Behavioral and Brain
Sciences, 22, 577–609 discussion 610–660.
Barsalou, L. W. (2008). Grounded cognition. Annual review of psychology, 59,
617–645.
Boroditsky, L. (2000). Metaphoric structuring: Understanding time through
spatial metaphors. Cognition, 75, 1–28.
Crawford, L. E. (2009). Conceptual metaphors of affect. Emotion Review, 1,
129–139.
Farias, A. R., Garrido, M., & Semin, G. R. (2013). Converging modalities
ground abstract categories: The case of politics. PloS One, 8(4), e60971.
Frost, P. (1988). Human skin color: A possible relationship between its sexual
dimorphism and its social perception. Perspectives in Biological Medicine,
32, 38–59.
Gibbs, R. W., Jr. (2006). Embodiment and cognitive science. New York:
Cambridge University Press.
Jablonski, N. G. (2004). The evolution of human skin and skin color. Annual
Review of Anthropology, 33(1), 585–623.
Jablonski, N. G., & Chaplin, G. (2000). The evolution of human skin
coloration. Journal of Human Evolution, 39(1), 57–106.
Jablonski, N. G., & Chaplin, G. (2002). Skin. Scientific American, 75–81.
Lakoff, G., & Johnson, M. (1980). Metaphors we live by. Chicago, IL: The
University of Chicago Press.
225
Lakoff, G., & Johnson, M. (1999). Philosophy in the flesh: The embodied mind
and its challenges to western thought. New York: Basic Books.
Lee, S. W. S., & Schwarz, N. (2011). Wiping the slate clean: Psychological
consequences of physical cleansing. Current Directions in Psychological
Science, 20, 307–311.
Meier, B. P., Robinson, M. D., Crawford, L. E., & Ahlvers, W. J. (2007). When
‘light’ and ‘dark’ thoughts become light and dark responses: Affect biases
brightness judgments. Emotion, 7, 366–376, http://dx.doi.org/10.1037/
1528-3542.7.2.366.
Meier, B. P., Robinson, M. D., & Clore, G. L. (2004). Why good guys wear
white: Automatic inferences about stimulus valence based on brightness.
Psychological Science, 15, 82–87.
Meier, B. P., Hauser, D. J., Robinson, M. D., Friesen, C. K., &
Schjeldahl, K. (2007). What’s ‘‘up’’ with God? Vertical space as a
representation of the divine. Journal of Personality and Social
Psychology, 93, 699–710.
Nestor, A., & Tarr, M. J. (2008). The segmental structure of faces and its use in
gender recognition. Journal of Vision, 8, 1–12.
Nestor, A., Vettel, J. M., & Tarr, M. J. (2008). Task-specific codes for face
recognition: How they shape the neural representation of features for
detection and individuation. PloS One, 3(12), e3978.
Niedenthal, P. M., Barsalou, L. W., Winkielman, P., Krauth-Gruber, S., & Ric,
F. (2005). Embodiment in attitudes, social perception, and emotion.
Personality and Social Psychology Review, 9, 184–211.
Schubert, T. (2005). Your highness: Vertical positions as perceptual symbols of
power. Journal of Personality and Social Psychology, 89, 1–21.
Semin, G. R., Garrido, M., & Farias, A. R. (in press). How many processes does
it take to ground a concept? In J. W. Sherman, B. Grabowski, & Y. Trope
(Eds.), Dual process theories of the social mind (in press).
Semin, G. R., Garrido, M. V., & Palma, T. A. (2012). Socially situated cognition:
Recasting social cognition as an emergent phenomenon. In S. Fiske, & N.
Macrea (Eds.), Sage handbook of social cognition (pp. 138–164). California:
Sevenoaks: Sage.
Semin, G. R., Garrido, M. V., & Palma, T. A. (2013). Interfacing body, mind,
the physical, and social world: Socially situated cognition. In D. E. Carlston
(Ed.), The Oxford handbook of social cognition (pp. 637–655). New York:
Oxford University Press.
Semin, G. R., & Smith, E. R. (2002). Interfaces of social psychology
with situated and embodied cognition. Cognitive Systems Research, 3,
385–396.
Semin, G. R., & Smith, E. (2013). Socially situated cognition in perspective.
Social Cognition, 31, 125–146.
Sherman, G. D., & Clore, G. L. (2009). The color of sin: White and black are
perceptual symbols of moral purity and pollution. Psychological Science,
20, 1019–1025.
Smith, E. R., & Semin, G. R. (2004). Socially situated cognition: Cognition in
its social context. Advances in Experimental Social Psychology, 36,
53–117.
Thibodeau, P. H., & Boroditsky, L. (2011). Metaphors we think with: the role
of metaphor in reasoning. PloS One, 6(2), e16782.
Wilson, M. (2002). Six views of embodied cognition. Psychonomic Bulletin
and Review, 9, 625–636.
Zhong, C., & Liljenquist, K. (2006). Washing away your sins: Threatened
morality and physical cleansing. Science, 313, 1451–1452.