Color Saturation Increases Perceived

Product Size

HENRIK HAGTVEDT
S. ADAM BRASEL

This research demonstrates a visual phenomenon with broad implications for con-
sumers: the perceived size of products depends on the saturation of their color.
Results from six experiments, employing objects and products with various
shapes and hues, show that increasing color saturation increases size percep-
tions. This influence is explained by the tendency for saturated color to capture at-
tention, which, in turn, is explained by the arousal that saturated color stimulates.
This research also demonstrates several downstream outcomes of the effect of
saturation on size perceptions: evaluations are more favorable—and willingness
to pay is higher—for products with high (low) saturation when usage goals call for
large (small) size. Additionally, participants choose more of a product to fill a con-
tainer with higher saturation. Further, the saturation of an object’s color affects the
perceived size of its surroundings, such that when a product with high (vs. low)
saturation is used as a benchmark, the environment is perceived to be compara-
tively smaller (vs. larger). Implications for aesthetics, design, sensory marketing,
and related topics are discussed. Lastly, to aid future color research, appendix A
outlines general challenges and recommendations in connection with the concep-
tualization, manipulation, and measurement of color.

Keywords: color, saturation, size, aesthetics, attention, arousal

C olor is ubiquitous. Recognizing its potential for di-

verse and far-reaching influences on consumer behav-
ior, recent literature has emphasized the need for more
2012). The current research demonstrates that the per-
ceived size of an object depends on the saturation of its
color. From a practical perspective, this research elucidates
research on color-related phenomena (Kareklas, Brunel, a broadly applicable device to influence consumer re-
and Coulter 2014; Labrecque and Milne 2012; Lee et al. sponses to products, environments, and marketing mate-
2014). One potential focus is suggested by literature ex- rials. From a theoretical perspective, our findings provide
ploring influences on consumers’ size perceptions new insights into fundamental effects of color on percep-
(Aydino glu and Krishna 2011; Ordabayeva and Chandon tion and judgment.
2013; Sevilla and Kahn 2014; Van Ittersum and Wansink Research typically distinguishes between three dimen-
sions of color (Hagtvedt and Brasel 2016; Labrecque,
Henrik Hagtvedt is an associate professor of marketing at the Carroll Patrick, and Milne 2013). While various models provide
School of Management, Boston College (Carroll School of Management, somewhat different conceptualizations of these dimen-
Boston College, Fulton Hall 450D, 140 Commonwealth Avenue, Chestnut sions—see appendix A for an expanded discussion on the
Hill, MA 02467; phone: (617) 552-4034; fax: (617) 552-6677; email: conceptualization, manipulation, and measurement of
[email protected]). S. Adam Brasel is an associate professor of marketing
color—we here rely on the common HSV (hue-saturation-
at the Carroll School of Management, Boston College (Carroll School of
Management, Boston College, Fulton Hall 443, 140 Commonwealth
value) model. Hue is the perceptual attribute corresponding
Avenue, Chestnut Hill, MA 02467; phone: (617) 552-1952; fax: (617) to a color’s dominant wavelength in the electromagnetic
552-6677; email: [email protected]). visible spectrum, resulting in our perceptions of red, green,
blue, and so on. Value refers to the color’s level of dark-
Darren Dahl served as editor and Rajesh Bagchi served as associate edi-
ness, with low value tending toward black. Saturation re-
tor for this article.
fers to the color’s purity, sometimes described as
Advance Access publication January 23, 2017 colorfulness relative to brightness. Whereas much of the

C The Author 2017. Published by Oxford University Press on behalf of Journal of Consumer Research, Inc.
V
All rights reserved. For permissions, please e-mail: [email protected]  Vol. 44  2017
DOI: 10.1093/jcr/ucx039

396
Downloaded from https://academic.oup.com/jcr/article-abstract/44/2/396/2939542
by guest
on 31 March 2018
 HAGTVEDT AND BRASEL 397

extant color research focuses on the effects of hue (Bagchi saturation and value on perceptions related to physical
and Cheema 2013; Mehta and Zhu 2009), saturation repre- properties such as size and weight. For example, objects
sents an equally important yet underinvestigated dimension with darker (vs. lighter) color appear heavier (Walker,
of color. Notably, saturation was investigated in only three Francis, and Walker 2010). Conversely, lighter color ap-
of 19 articles covered in a recent review of color research pears to increase size perceptions as compared to their
in marketing (Labrecque et al. 2013). darker counterparts (Mahnke 1996). For instance,
Across six experiments using various stimuli and presen- Gundlach and Macoubrey (1931) demonstrate that light ob-
tation techniques, we demonstrate that consumers perceive jects appear comparatively larger than dark ones, while
objects with highly saturated color as larger than otherwise Nakatani (1989) similarly shows that lighter circles are
identical objects with less saturated color of the same hue perceived to be larger than darker ones. Curiously, one part
and value. We further propose and empirically support that of a study in an early article from the psychology literature
this effect arises because highly saturated color captures suggests a link between saturation and perceived size as
consumers’ attention due to the arousal it elicits. In other well, but it indicates a negative rather than positive correla-
words, more (vs. less) arousing, attention-grabbing stimuli tion (Tedford, Bergquist, and Flynn 1977; see the General
are perceived to be larger. Additionally, we illustrate Discussion section). However, the influence of saturation
downstream consequences of the saturation–size effect: on perceived size has yet to be systematically investigated;
study participants exhibit more favorable evaluations and it remains a fundamental gap in knowledge about visual
higher willingness to pay for products with high (low) satu- perception. Relatedly, Labrecque et al. (2013) acknowl-
ration when usage goals call for large (small) size. Further, edge the paucity of prior color work and explicitly call for
participants choose more of a product to place in a con- researchers to investigate whether the color of a package
tainer with high (vs. low) saturation. Finally, the saturation can affect its apparent size.
of a product’s color also affects the perceived size of its
surroundings, such that when a product with high (vs. low) The Influence of Color Saturation on Arousal,
saturation is used as a benchmark, the environment is per-
ceived to be comparatively smaller (vs. larger).
Attention, and Perceived Size
In the current research, we hypothesize that objects with
highly saturated color are perceived to be larger than those
THEORETICAL FRAMEWORK with less saturated color. We base this prediction on the
notions that a salient, intense stimulus is likely to generate
Consumer Responses to Color arousal and attract attention, and an attention-grabbing
Despite the ubiquity of color and its widespread poten- stimulus is likely to appear large. Whereas larger elements
tial to influence consumers, relatively few consumer re- attract increased visual attention (Pieters and Wedel 2004),
searchers have focused on this topic. Nonetheless, some this mental link can work both ways, such that elements
notable cognitive and affective responses to color hue have that attract increased visual attention appear larger (Folkes
been documented, ranging from brand personality percep- and Matta 2004). This seems especially likely given that
tions to the moral acceptability of undesirable behaviors the capture of attention can affect perceptual experience as
(De Bock, Pandelaere, and Van Kenhove 2013; Labrecque well as basic visual processing (Gobell and Carrasco
and Milne 2012). In connection with ad photos, the persua- 2005). In other words, the relationship between size and at-
sive impact of color depends on the correspondence be- tention may work both ways, such that a stimulus that cap-
tween available and required processing resources tures more attention is perceived to be larger than an
(Meyers-Levy and Peracchio 1995). Further, black-and- objectively equal-sized stimulus that captures less
white (vs. color) imagery is cognitively associated with attention.
high-level (vs. low-level) construal (Lee et al. 2014). In re- The notion that increased saturation leads to increased
lated work, Puccinelli et al. (2013) demonstrate that male attention is generally supported by research establishing
consumers perceive greater savings when prices are pre- that salient or intense stimuli capture visual attention
sented in red rather than black. When investigating cogni- (Michael and Galvez-Garcıa 2011; Mizzi and Michael
tive task performance, Mehta and Zhu (2009) find that blue 2014). While increased visual attention leads to increased
(red) evokes approach (avoidance) motivation and en- perceptions of saturation (Fuller and Carrasco 2006), there
hances performance on a creative (detail-oriented) task. A is also evidence that this relationship is bidirectional; prior
blue (vs. red) background color also elicits higher price of- work provides preliminary support that saturated color cap-
fers in negotiations but lower bid jumps in online auctions; tures attention, regardless of the background color against
the effects arise because red encourages aggression by in- which it is contrasted (Camgöz, Yener, and Güvenç 2004).
ducing arousal (Bagchi and Cheema 2013). Relatedly, saturation has been used as a salience-enhancing
In research closer to the current undertaking, scholars of manipulation to aid visual search (Braun 1994). Saturation
visual psychology have investigated the influence of color is a common input dimension in computer vision programs

Downloaded from https://academic.oup.com/jcr/article-abstract/44/2/396/2939542

by guest
on 31 March 2018
 398 JOURNAL OF CONSUMER RESEARCH

and neurological vision models that predict human visual the laptop elicits arousal and grabs their attention.
attention (Aziz et al. 2006; Frey, Honey, and König 2008), Mediation analysis demonstrates a three-step path in which
and increasing saturation also appears to increase target- saturation increases size perceptions via arousal and atten-
ground contrast (Dresp-Langley and Reeves 2014; Egusa tion. The final three studies were designed to investigate
1983), which in turn would increase attentional capture for downstream consequences of the saturation–size relation-
highly saturated objects. ship. In study 4, when a usage goal calls for a large (small)
These extant findings are consistent with common obser- size, a carry-on suitcase with high (low) color saturation
vations in nature regarding flowers or venomous animals encourages more favorable evaluations and higher willing-
that capture attention because they are adorned with in- ness to pay. In study 5, participants serve themselves a
tense color. Notably, an evolutionary argument supports larger quantity of jelly beans when filling cups with more
the connection between intense color and both attention highly saturated color, consistent with the expectation that
and arousal, whether the color is associated with reward or the more saturated cups are perceived as larger. Finally,
danger. It would have benefited our ancestors if both ripe study 6 investigates whether an object’s color saturation af-
fruit and venomous animals aroused them and attracted fects the perceived size of its surroundings; the ceiling
their attention. This observation also fits with existing find- height of a room is estimated to be comparatively lower
ings that saturated color elicits excitement in contexts such when an ottoman with high (vs. low) color saturation is
as advertising (Gorn et al. 1997; Gorn et al. 2004; used as a benchmark. See figure 1 for a diagrammatical
Labrecque and Milne 2012). Further, heightened arousal overview.
induces alertness and is associated with focusing attention
on the arousing stimulus (Adolphs and Damasio 2001;
Gable and Harmon-Jones 2008; Ohman,€ Flykt, and Esteves STUDY 1
2001). In sum, we hypothesize that highly saturated color Fifty-six undergraduates (70% male) participated in a
elicits arousal, thereby capturing attention, thereby encour- within-subjects experiment in return for snacks with a
aging increased size perceptions. Formally stated: value of approximately $1. In groups, participants were
H1: An object with highly saturated color is perceived to be shown a pair of green cubes projected on a large screen,
larger than an otherwise identical object with compara- which served as the basis for completing paper question-
tively less saturated color of the same hue and value. naires. The left-right position of the cubes was counterbal-
H2: An object with highly saturated color captures more at- anced to avoid any location-specific influence. Shape, size,
tention than an otherwise identical object with compar- hue, and color value were equivalent across the two cubes,
atively less saturated color of the same hue and value. but they were high versus low in saturation. In this study—
H3: The influence of color saturation on perceived size fol-
and in the other studies relying on electronic stimuli—the
lows a three-step mediational pathway in which satura- color measures were calibrated using Adobe Photoshop.
tion influences arousal, which influences attention, See appendix B for color specifications and appendix C for
which influences perceived size. stimuli. In a forced-choice procedure, participants reported
which of the two cubes appeared to be larger.
A subsequent chi-square test revealed that 37 (66.1%) of
the participants perceived the high-saturation cube to be
EMPIRICAL OVERVIEW larger versus 19 (33.9%) who perceived the low-saturation
cube to be larger (v2(1) ¼ 5.79, p < .05). These results
We test our hypotheses with a set of six experiments, support hypothesis 1.
some of which utilize geometric primitives (i.e., 3D The following study was designed to replicate the influ-
shapes) and some of which utilize consumer products. Hue ence of saturation on size perceptions while implicating at-
and value are held constant within each experiment, but we tention as the process mechanism.
include several hues across the experiments for the sake of
generalizability. In study 1, participants are exposed to
cubes with high- versus low-saturation color and identify STUDY 2
which one they perceive to be larger. In study 2, partici-
pants estimate the size of geometric figures with high- ver-
Procedure
sus low-saturation color, while attention is measured via Fifty-eight undergraduates (55% female), sourced via
eye tracking. This measure mediates the influence of satu- on-campus recruitment, participated in an experiment and
ration on perceived size. Additionally, pupil dilation mea- were compensated with a $5 gift card to a major online re-
sures suggest that objects with higher saturation elicit tailer. Participants entered the lab individually and, after
marginally increased arousal. In study 3, participants esti- completing informed consent, were seated and calibrated
mate the screen size of a laptop computer with high- versus with a traditional nine-point calibration field to an ASL
low-saturation color. They also report the extent to which 6000 desk-mounted corneal-reflection eye tracker. This

Downloaded from https://academic.oup.com/jcr/article-abstract/44/2/396/2939542

by guest
on 31 March 2018
 HAGTVEDT AND BRASEL 399

FIGURE 1

OVERVIEW OF THE EMPIRICAL INVESTIGATION

Color saturation: Arousal Attention Perceived size Downstream

High versus Low consequences

Studies 1–6 Studies 2 and 3 Studies 2–4 Studies 1–3 Studies 4–6

tracker allowed gaze capture at 60 frames per second; dif- versus 40.5% for the low-saturation object (v2(1) ¼ 4.17,
ferent fixation floors from 150 to 250 milliseconds were p < .05). These results support hypothesis 1.
explored with no significant effect on results, so a com-
Attention. For single-object exposures, the amount of
monly used 175 millisecond fixation floor for results is re-
time fixated on the object (vs. the background) was higher
ported here. Participants were exposed to a series of
for high-saturation objects versus low-saturation objects
images, with verbal questions after each stimulus exposure;
(M ¼ 3.06 vs. M ¼ 2.71, t(54) ¼ 2.10, p < .05, Cohen’s
stimulus onset was triggered by the experimenter via a
d ¼ .44). For paired-object exposures, the high-saturation
wireless mouse, and the program automatically removed
object was first attended to 60.3% of the time, versus
the stimulus from the screen after the 4 second duration.
39.7% for the low-saturation object (v2(1) ¼ 4.97, p <
All stimuli were presented against white backgrounds.
.05). The high-saturation object also captured significantly
Participants were first exposed for 4 seconds to a pair of 6
more overall fixation time than the low-saturation object
centimeter objects onscreen (cubes or spheres), either red
(2.25 seconds vs. 1.63 seconds, paired t(54) ¼3.87, p <
or blue, where one object had high saturation and the other
.01, Cohen’s d ¼ .67). These results support hypothesis 2.
object had low saturation. After this, they were exposed for
4 seconds to a single red or blue object (cube or sphere), Mediation. To test whether visual attention mediated
presented in isolation off-center on the screen, which had the effect of saturation on size estimates, bootstrap estima-
either high or low saturation. This was followed by a pair tion with 10,000 resamples was conducted on the singleton
of objects in the shape and color not seen previously, and a exposure conditions (PROCESS model 4; Hayes 2013)
final single object in the shape and color not seen previ- with saturation as the independent variable, benchmarked
ously. See appendix B for specifications and appendix C size estimates as the dependent variable, and fixation dura-
for stimuli. Order of shape and color was randomized tion on the object as the mediator. The indirect effect was
across participants; object shape and hue did not have a significant (B ¼ .24, SE ¼ .13; 95% CI ¼ .03, .54), indicat-
significant effect on results so are not discussed further. ing mediation. These results are consistent with hypothesis
For single-object exposures, the eye tracker recorded the 3. Additionally, the more saturated object in the paired-
duration of fixations on the object. Participants were then object scenarios was identified as larger 69% of the time
verbally asked to estimate the diagonal size of the object in when it was attended to first, but only 50% of the time
centimeters between 2 and 10 centimeters, after being told when it was attended to second.
that the average object in the study was 6 centimeters (to
Arousal. As a preliminary step in implicating arousal
serve as an initial anchor). For paired-object exposures, the
in the effect of saturation on attention and perceived size,
eye tracker recorded which object was first fixated on and
we investigated pupil dilation during stimulus exposure
the duration of fixations on each object, and the experi-
(Bradley et al. 2008). As there exists considerable variation
menter verbally asked the participants which object they
in pupil diameter across individuals due to gender, body
felt was the larger of the two.
size, and other random factors (McLachlan and Howland
2002), each participant’s pupil diameter was normalized
against the participant average during the calibration pro-
Results cedure. If this normalization point is set to 1, pupil diame-
Size Estimates. For single-object exposures, high- ter during stimulus exposure ranged from .83 to 1.25. Note
saturation objects were rated as significantly larger than that the initial 2 seconds of each stimulus exposure are not
low-saturation objects (M ¼ 6.40 centimeters vs. M ¼ 5.71 used, given the reflexive shrinking of the pupil due to the
centimeters, t(54) ¼ 3.02, p < .01, Cohen’s d ¼ .61). For initial luminance shift; only pupil dilation from the second
paired-object exposures, participants chose the highly satu- half of each exposure is utilized. Pupil dilation was mar-
rated object as the larger of the two 59.5% of the time, ginally increased for the more (vs. less) saturated objects

Downloaded from https://academic.oup.com/jcr/article-abstract/44/2/396/2939542

by guest
on 31 March 2018
 400 JOURNAL OF CONSUMER RESEARCH

when mean dilation across all stimuli was examined (.96 participant self-reported color blindness, and neither this
vs. .93, paired t(56) ¼ 1.94, p < .06). In addition, pupil di- measure nor gender was a significant covariate or signifi-
lation was marginally greater for the more (vs. less) satu- cantly altered the results, so neither is discussed further.
rated object in the paired-object presentations (.96 vs. .93,
t(56) ¼ 1.76, p < .10).
Note that pupil dilations due to changes in arousal are Results
much smaller than dilations due to changes in luminance Size Estimate. An ANOVA with color saturation as the
(Beatty and Lucero-Wagoner 2000); one would not expect independent variable and size estimate as the dependent
a large effect from saturation shifts alone. Still, the mar- variable revealed that the laptop with high (vs. low) satura-
ginal effect is consistent with our expectation that high sat- tion was perceived to be larger (Mhigh ¼ 15.28, SD ¼ 1.26
uration triggers arousal, in addition to attention and vs. Mlow ¼ 14.41, SD ¼ 1.59, F(1, 75) ¼ 7.13, p < .01, gp2
increased size perceptions. ¼ .09). This result supports hypothesis 1.

Discussion Attention. A similar ANOVA on the attention scale re-

vealed that the laptop with high (vs. low) saturation cap-
Study 2 provides further evidence supporting the pre-
tured more attention (Mhigh ¼ 5.19, SD ¼ 1.08 vs. Mlow ¼
dicted positive effect of saturation on perceived size and
4.66, SD ¼ .98, F(1, 75) ¼ 4.95, p < .05, gp2 ¼ .06). This
also implicates attention as the process mechanism under-
result supports hypothesis 2.
lying this effect. Further, preliminary pupil dilation evi-
dence suggests a potential role for arousal in this effect.
Arousal. A similar ANOVA on arousal revealed that
The next study was designed to investigate the full media-
the laptop with high (vs. low) saturation stimulated higher
tional chain.
arousal (Mhigh ¼ 6.05, SD ¼ 1.34 vs. Mlow ¼ 5.14,
SD ¼ 1.62, F(1, 75) ¼ 7.35, p < .01, gp2 ¼ .09).
STUDY 3
Mediation. To test whether visual attention mediated
Procedure the effect of saturation on size estimates, bootstrap estima-
Eighty undergraduates participated in a between- tion (PROCESS model 4; Hayes 2013) with 10,000 resam-
subjects experiment in return for snacks with a value of ap- ples was conducted with saturation as the independent
proximately $1. In groups, participants were shown an variable, size estimate as the dependent variable, and atten-
image of a red laptop projected on a large screen, which tion as the mediator. The indirect effect was significant
served as the basis for completing paper questionnaires. (B ¼ .36, SE ¼ .19; 95% CI ¼ .07, .80), indicating media-
The laptop was either high or low in color saturation; see tion, whereas the direct effect became nonsignificant (B ¼
appendix B for color specifications and appendix C for .51, SE ¼ .29; 95% CI ¼ –.08, 1.09).
stimuli. Three participants did not complete the question- To test whether arousal mediated the effect of saturation
naires, so responses from seventy-seven participants (49% on attention, a similar bootstrap estimation was conducted
male) were included in the analysis. with saturation as the independent variable, attention as the
Based on their visual impression, participants estimated dependent variable, and arousal as the mediator. The indi-
the laptop’s screen size on a scale with 1 inch increments, rect effect was significant (B ¼ .41, SE ¼ .16; 95% CI ¼
ranging from 10 to 20 inches. They next reported, using .13, .77), indicating mediation, whereas the direct effect
seven-point scale items (1 ¼ Not at all, 7 ¼ Definitely), the became nonsignificant (B ¼ .12, SE ¼ .19; 95% CI ¼ –.26,
extent to which they agreed that “The laptop is attention- .50).
getting” and “The laptop captures my attention” (later Finally, to test the full mediational pathway, a similar
combined in an attention scale; r ¼ .60). Participants re- bootstrap estimation (PROCESS model 6; Hayes 2013)
ported their arousal and mood using the affect grid (Kupor was conducted with saturation as the independent variable,
and Tormala 2015; Russell, Weiss, and Mendelsohn 1989), size estimate as the dependent variable, and arousal and at-
which is a 9  9 matrix that treats arousal and mood as or- tention as successive mediators. The total indirect effect
thogonal dimensions; it comprises two nine-point scales, was significant (B ¼ .53, SE ¼ .21; 95% CI ¼ .16, 1.01),
simultaneously providing measures for arousal (low–high) indicating mediation, whereas the direct effect became
and mood (displeasure–pleasure). No differences were ex- nonsignificant (B ¼ .34, SE ¼ .28; 95% CI ¼ –.22, .90).
pected or found for mood (M ¼ 5.43 vs. M ¼ 5.49, p > More specifically, the indirect effect via both mediators
.87), so this variable is not discussed further. Participants was significant (B ¼ .14, SE ¼ .08; 95% CI ¼ .03, .36), as
also reported gender and whether they were color-blind; al- was the indirect effect via arousal alone (B ¼ .35, SE ¼
though the current research focuses on saturation rather .18; 95% CI ¼ .08, .77), whereas the indirect effect via at-
than hue, we deemed it feasible that color blindness might tention alone became nonsignificant (B ¼ .04, SE ¼ .08;
influence the effects under investigation. Only one 95% CI ¼ –.07, .27). These results support hypothesis 3.

Downloaded from https://academic.oup.com/jcr/article-abstract/44/2/396/2939542

by guest
on 31 March 2018
 HAGTVEDT AND BRASEL 401

Discussion benchmark served to avoid overly large amounts of vari-

ance. Next, participants reported attention on the same
These results replicated those from the prior two studies,
scale items as in the previous study (r ¼ .79). Finally, par-
demonstrating the influence of color saturation on the per-
ticipants reported color blindness and gender. Five partici-
ceived size of consumer products (laptops). They also rep-
pants self-reported color blindness, but neither this
licated the process evidence from study 2, using a different
measure nor gender was a significant covariate or signifi-
measure; whereas study 2 relied on eye tracking, study 3
cantly altered the results, so they are not discussed further.
relied on self-reported attention. Additionally, the results
from study 3 illustrated the role of arousal in the influence
of saturation on attention and demonstrated the full media- Results
tional pathway from saturation to size estimate via arousal Evaluation. An ANOVA with saturation and goal as
and attention. Having thus provided evidence for the novel the independent variables and evaluation as the dependent
effect of saturation on perceived size, and having illumi- variable revealed a main effect of goal (Mlarge ¼ 3.80,
nated the underlying process, we investigate in our next SD ¼ 1.57 vs. Msmall ¼ 4.55, SD ¼ 1.29, F(1, 152) ¼
three studies the downstream consequences of this effect. 10.30, p < .005, gp2 ¼ .06) and a saturation  goal interac-
tion (Mlarge goal, high saturation ¼ 3.97, SD ¼ 1.57 vs.
Mlarge goal, low saturation ¼ 3.65, SD ¼ 1.57 vs. Msmall goal, high saturation
STUDY 4 ¼ 4.14, SD ¼ 1.25 vs. Msmall goal, low saturation ¼ 4.92,
Study 4 was designed to investigate the influence of SD ¼ 1.23, F(1, 152) ¼ 5.86, p < .05, gp2 ¼ .04). See
color saturation on product evaluation and willingness to figure 2. There were no other significant effects. Contrast
pay, depending on usage goals. We expected a product analysis revealed that when the goal was a large size, the
with high (vs. low) saturation to fit with a goal that calls carry-on with high (vs. low) saturation was evaluated
for a large (vs. small) product size. directionally more favorably (p ¼ .34), whereas when the
goal was a small size, the carry-on with low (vs. high)
saturation was evaluated significantly more favorably
Procedure (p < .05).
One hundred sixty undergraduates participated in a Willingness to Pay. A similar ANOVA on willingness
between-subjects experiment in return for snacks with a to pay revealed a main effect of goal (Mlarge ¼ 51.23,
value of approximately $1. In groups, participants were SD ¼ 21.82 vs. Msmall ¼ 58.98, SD ¼ 22.83, F(1, 152) ¼
shown an image of an orange carry-on suitcase projected 4.30, p < .05, gp2 ¼ .03) and a marginally significant satu-
on a large screen, which served as the basis for completing ration  goal interaction (Mlarge goal, high saturation ¼ 52.43,
paper questionnaires. The carry-on was either high or low SD ¼ 25.96 vs. Mlarge goal, low saturation ¼ 50.13, SD ¼ 17.46
in color saturation; see appendix B for color specifications vs. Msmall goal, high saturation ¼ 53.72, SD ¼ 25.18 vs.
and appendix C for stimuli. Four participants did not com- Msmall goal, low saturation ¼ 63.64, SD ¼ 19.66, F(1, 152) ¼
plete the questionnaire, so responses from 156 participants 2.93, p < .10, gp2 ¼ .02). There were no other significant
(53% female) were included in the analysis. In the ques- results. Contrast analysis revealed that when the goal was a
tionnaire, participants were given a scenario that stipulated large size, participants were willing to pay directionally
the consumer goal of a large or small carry-on, depending more for the carry-on with high (vs. low) saturation (p ¼
on a randomly assigned condition: “Imagine that you want .66), whereas when the goal was a small size, participants
to purchase a carry-on suitcase that is large enough to fit were willing to pay significantly more for the carry-on
plenty of your belongings” (vs. “small enough to fit easily with low (vs. high) saturation (p < .05).
into an overhead storage compartment”). We expected par-
Attention. A similar ANOVA on the attention scale re-
ticipants with the goal of a large (vs. small) carry-on to pre-
vealed that the carry-on with high (vs. low) saturation cap-
fer the one with high (vs. low) color saturation, given the
tured more attention (Mhigh ¼ 4.97, SD ¼ 1.38 vs. Mlow ¼
hypothesized influence of color saturation on perceived
3.70, SD ¼ 1.55, F(1, 152) ¼ 28.03, p < .0005, gp2 ¼ .16).
size. The study was a 2 (saturation: high vs. low)  2
There were no other significant results.
(goal: large vs. small) between-subjects experiment.
Based on their given goal, participants reported their Mediation. To test whether visual attention mediated
evaluation of the carry-on using seven-point scale items the moderated effect of saturation on evaluation, bootstrap
(unfavorable–favorable, negative–positive, bad–good, un- estimation (PROCESS model 5; Hayes 2013) with 10,000
pleasant–pleasant, dislike very much–like very much), later resamples was conducted with saturation as the indepen-
combined in an evaluation scale (a ¼ .94). Next, partici- dent variable, evaluation as the dependent variable, goal as
pants were asked, “If the average price of a carry-on suit- the moderator, and attention as the mediator. The indirect
case is $75, how much would you be willing to pay for this effect was significant (B ¼ .43, SE ¼ .12; 95% CI ¼ .23,
one?” Because the question was open-ended, the $75 .70), indicating mediation.

Downloaded from https://academic.oup.com/jcr/article-abstract/44/2/396/2939542

by guest
on 31 March 2018
 402 JOURNAL OF CONSUMER RESEARCH

FIGURE 2

THE INFLUENCE OF COLOR SATURATION ON PRODUCT EVALUATION

5.5
3.97 3.65
4.14 4.92
5
Product evaluation

4.5

3.5

2.5
Large goal Small goal

High saturation Low saturation

Discussion Procedure
Study 4 extended the findings from the previous studies Ninety-one students (51% male) participated, on an indi-
to incorporate the role of consumer goals. The pattern of vidual, voluntary basis, in a study designed to investigate
results largely supported our expectations, with high (vs. behavioral responses to color saturation. Solid-color, 9
low) saturation being preferred among participants with a ounce, heavy-duty paper cups had been custom-made by a
large-size goal and low (vs. high) saturation being pre- professional printing service for the study as stimuli with
ferred among participants with a small-size goal. The for- high versus low saturation; see appendix B for color speci-
mer effect was only directional, perhaps because the fications and appendix D for stimuli. A large, clear plastic
shape of the carry-on indicated a smaller size, thereby jar was filled with mixed Jelly Belly jelly beans. The ratio-
limiting the capacity for saturation to signal a larger size. nale for choosing jelly beans was twofold: first, we wanted
This observation serves as a reminder that color saturation a food product that was easily divisible into small units, as
is only one of several design decisions, and saturation this would allow participants to control the quantity of their
alone may not be enough to override responses to, for in- servings. Second, we assumed that the wide variety of col-
stance, product shape. However, the results suggest that ors in a jelly bean mix was less likely than a single, spe-
saturation is one of the factors contributing to perceptions cific color to interact with the color of the cup and possibly
of how well a given product aligns with consumers’ size- skew the results.
related goals. Additionally, study 4 provided some limited The jar with jelly beans was placed on a table covered
evidence that willingness to pay is similarly affected by by a grey tablecloth. Under the tablecloth, directly beneath
saturation. The following study was designed to investi- the jar, was an electronic food scale. Any changes in the
gate whether the influence of saturation also extends to amount of jelly beans would therefore be registered by the
behavior. scale on which the jar was sitting. A pull-out display, con-
nected via wire to the scale, was extended to be visible
from beneath one end of the tablecloth; the experimenter
STUDY 5 could thereby track any changes in the weight registered
by the scale. A 6 ounce, stainless steel ladle was placed
Study 5 was designed to investigate whether the color next to the jar. The rationale for using a ladle with a
saturation of a recipient container influences the amount of smaller capacity than that of the cups was that a rounded
jelly beans that consumers serve themselves. We expected scoop would be approximately sufficient to fill a cup to the
that study participants would serve themselves a larger rim.
quantity if the saturation of the container’s color was high Participants were told that they should take one scoop of
(vs. low), because high (low) saturation indicates large jelly beans to put in their cup and that they could take as
(small) size. much or as little as they wanted. In case some participants

Downloaded from https://academic.oup.com/jcr/article-abstract/44/2/396/2939542

by guest
on 31 March 2018
 HAGTVEDT AND BRASEL 403

did not want to eat jelly beans, they were also told that would you estimate that the ceiling is?” They gave their
they could do whatever they wanted with the jelly beans open-ended response in inches and then reported color
they served themselves. Participants were randomly as- blindness and gender. Two participants reported being
signed to a cup with either high or low saturation. As a par- color-blind, but neither this measure nor gender was a sig-
ticipant used the ladle to take a scoop of jelly beans, the nificant covariate.
experimenter recorded the weight reduction in grams. Our
expectation was that high (vs. low) color saturation would Results and Discussion
cause the cup to be perceived as larger (vs. smaller), which
would prompt participants to take a larger (vs. smaller) As expected, an ANOVA with saturation as the indepen-
scoop of jelly beans to fill the cup (Wansink 2006). dent variable and perceived ceiling height as the dependent
variable revealed that estimates based on the high-
saturation ottoman were lower than estimates based on the
Results and Discussion low-saturation ottoman (Mhigh ¼ 128.04, SD ¼ 26.51 vs.
An ANOVA with saturation as the independent variable Mlow ¼ 151.61, SD ¼ 50.16, F(1, 48) ¼ 4.50, p < .05, gp2
and grams of jelly beans as the dependent variable revealed ¼ .09). Although gender and color blindness did not exert
that participants with high-saturation cups served them- a significant influence when added as covariates, they re-
selves more jelly beans than did participants with low- duced the effect of saturation to marginal significance
saturation cups (Mhigh ¼ 88.58, SD ¼ 37.83 vs. Mlow ¼ (F(1, 46) ¼ 3.72, p ¼ .06, gp2 ¼ .08).
69.80, SD ¼ 34.84, F(1, 89) ¼ 6.07, p < .05, gp2 ¼ .06), These results provide evidence that an object’s color sat-
consistent with the expectation that participants would uration can also influence how consumers perceive the en-
view the more saturated cups as larger. Perhaps surpris- vironment. If high (low) saturation causes an object to be
ingly, there was no effect of gender when added as a perceived as larger (smaller), and that object is used as a
covariate. benchmark by which to assess the surroundings, then those
The result of this study is arguably particularly striking, surroundings appear comparatively smaller (larger) as a
given the clear differences in behavior based on the cup’s result.
color saturation. Participants with high-saturation cups
served themselves on average 27% more than participants GENERAL DISCUSSION
with low-saturation cups, which is notable from a practical
consumer marketing perspective. Practical and Theoretical Contributions
The following, final study was designed to investigate Six studies support our prediction that highly saturated
another potential downstream consequence of color satura- color increases the perceived size of objects, as compared
tion: if consumers know the size of a given object, and to the same objects with less saturated color. The studies
they perceive that object to be relatively large (vs. small), incorporate various shapes and color hues, rely on both
this suggests that the environment will appear compara- geometric figures and consumer products, and use both
tively small (vs. large) as a result. physical and digital stimuli. Further, our results show that
our hypothesized effect arises because saturated color cap-
STUDY 6 tures attention, which in turn is due to the arousal it elicits:
the mediating role of attention is demonstrated with eye-
Study 6 was designed to investigate whether an object’s tracking measures as well as self-reported measures, and
color saturation affects the perceived size of its surround- the role of arousal is illustrated via three-step mediation
ings. We expected that a room’s ceiling height would ap- analysis. Additionally, we demonstrate downstream out-
pear comparatively lower when the color of a benchmark comes of the effect of saturation on size perceptions: evalu-
in the room—an ottoman—is high (vs. low) in saturation. ations are more favorable, and willingness to pay is higher,
for products with high (low) saturation when usage goals
call for large (small) size. Further, participants choose
Procedure more of a product to fill a container with high (vs. low) sat-
Fifty undergraduates (56% female) participated in a uration. Finally, the saturation of a product’s color also af-
between-subjects experiment in return for snacks with a fects the perceived size of its surroundings, such that when
value of approximately $1. In groups, participants were a product with high (vs. low) saturation is used as a bench-
shown an image of a grey room with a violet ottoman proj- mark, the environment is perceived to be comparatively
ected on a large screen, which served as the basis for com- smaller (vs. larger).
pleting paper questionnaires. The ottoman was either high In terms of theoretical contribution, this is the first re-
or low in color saturation; see appendix B for color specifi- search to demonstrate the positive relationship between
cations and appendix D for stimuli. Participants were color saturation and size perceptions. Additionally, we il-
asked: “If the violet ottoman is 20 inches tall, how high lustrate a full causal pathway that ties together saturation,

Downloaded from https://academic.oup.com/jcr/article-abstract/44/2/396/2939542

by guest
on 31 March 2018
 404 JOURNAL OF CONSUMER RESEARCH

arousal, attention, and size perceptions. Whereas prior re- interact with color effects (Hagtvedt and Brasel 2016;
search has focused on one or two of these variables in iso- Spence 2011). In general, the interplay of color perception
lation, the current research is the first to provide a more with other sensory modalities remains largely unexplored
holistic analysis. Additionally, the various downstream in the literature and could represent a fruitful area for fur-
outcomes highlight the evaluative and behavioral influ- ther exploration.
ences that may arise from a mere adjustment of saturation. The relationship between the process measures of
These findings also have a very broad practical relevance. arousal and attention is also in need of further investiga-
For example, color saturation is a low-cost way to influ- tion. The current state of scholarship establishes a connec-
ence attention and size perceptions in retail contexts. It tion between the two, but it remains an open question
may be even more critical in online contexts, given that the whether arousal precedes attention or vice versa, and in-
visual impression can be the only sensory gauge available deed it seems likely that each affects the other (Adolphs
to consumers. and Damasio 2001). Although we hypothesized and empir-
ically supported a process connecting saturation and size
via arousal first and attention second, an alternative path-
Limitations and Future Research way—with attention leading to arousal—is also supported
A general limitation of color research is the difficulty of by the data (see appendix E). The technology required to
ensuring that experimental stimuli reflect the exact attrib- conclusively answer this question of mediational order
utes specified. Electronic presentation varies depending on may not even exist yet, given the need to combine ex-
factors such as software and screens, and even the appear- tremely rapid measurement with rich information. To fur-
ance of tangible objects such as Munsell color chips or ther complicate matters, arousal and affective valence may
consumer products can change based on factors such as in- interact to influence the processing of items in visual atten-
terior lighting. Therefore, individual findings, including tion (McConnell and Shore 2011). Another question is the
the ones in the current research, should be interpreted with extent to which color effects rely on an automatic or con-
care. The multiple replications assuage such concerns to scious process. Future work investigating moderators of
some extent, but our investigations revealed nonsignificant color effects in consumer behavior will likely need to dis-
findings as well, especially for studies conducted online, tinguish between these types of processes.
presumably due in part to differences in the environment Another potential topic for investigation pertains to color
or in software and hardware used by participants. Not only influences in a changing environment. For instance, mar-
are type II errors common when studies rely on subtle ma- keters may be able to rely on saturation to guide visual at-
nipulations of this kind, but more research is also needed to tention in dynamic media contexts such as video. A question
uncover possible moderators. then arises as to how shifts in saturation might affect assess-
Notably, our findings consistently diverge from the one ments such as size estimates. Would an object appear to be
reported by Tedford et al. (1977). In preliminary studies, growing in size, or approaching the camera, if its color satu-
we tested whether it was possible to reverse our effect, but ration begins at a low level but gradually increases?
a reversal never occurred. While we can only speculate Likewise, would decreasing saturation lead to perceptions of
about the discrepancy between their result and ours, one shrinking or receding? If found, such effects could represent
possibility is a difference in how saturation was defined or additional ways for advertisers to encode meaning into ani-
operationalized (consistent with the HSI model they used; mated or video ad content, making saturation shifts effective
see appendix A). According to Elliot and Maier (2014), a in animated-limited commercial environments such as on-
common concern with most color research conducted be- line display advertising (Brasel and Hagtvedt 2016).
fore the 21st century was variability in color stimuli due to Interestingly, visual illusions do not always extend into
paper-based or slide-based presentation, a lack of total con- the tactile realm. For example, the aforementioned research
trol over the other two color dimensions when manipulat- by Walker et al. (2010) indicates that bigger and darker ob-
ing a target dimension such as saturation, and difficulty in jects look heavier, but feel lighter, than otherwise smaller
truly matching colors across conditions or experiments, and less dark objects. This suggests that there may arise a
leading to numerous contradictory and inconsistent find- disconnect between the perceptual saturation–size effect
ings in the literature. However, Tedford et al. appear to and corresponding tactile information when consumers
have shown great attention to detail, and they relied on physically handle an object. Indeed, even the specific man-
Munsell color chips, which arguably provide more consis- ner in which an object is handled may affect how it is per-
tent measures of color dimensions than most electronic ceived (e.g., arm flexion vs. extension; Van den Bergh,
methods do. Perhaps social or situational cues in the envi- Schmitt, and Warlop 2011). Future research could investi-
ronment moderated the effect, in which case it would be gate how such issues are resolved in the consumer’s mind,
beneficial to identify the moderating variable(s). Relatedly, as well as how the same visual cues may influence differ-
the growing literature on crossmodal correspondences ent assessments (e.g., size and weight/density)
highlights the potential for nonvisual sensory cues to simultaneously.

Downloaded from https://academic.oup.com/jcr/article-abstract/44/2/396/2939542

by guest
on 31 March 2018
 HAGTVEDT AND BRASEL 405

Future research could also map out potential contrast ef- how to compare effects of different dimensions of color.
fects, or the effects arising from various color combina- For instance, one can compare the effect of 50% saturation
tions. For example, a visual environment may contain with the effect of 75% saturation, but how can either one
different levels of saturation simultaneously, along with be compared with the effect of a 360 degree (i.e., red) hue?
different hues and different color values as well. Colors are Related research could focus on interaction effects be-
seldom presented in isolation, and very little research has tween color and other variables, including physical cues
so far focused on combined color effects (Deng, Hui, and such as product shape. In the current research, the satura-
Hutchinson 2010). Because highly saturated colors are tion–size effect appeared to operate similarly across differ-
more intense than less saturated ones, their contrast with ent shapes, but this may not generalize to all shapes or to
the environment is arguably often more pronounced. Prior all color effects. Alternatively, future work could examine
work has demonstrated that contrast may play a role in size the role of psychological factors such as the perceiver’s
perceptions. For example, Van Ittersum and Wansink mindset or, conversely, investigate how color influences
(2012) show that color contrast influences the Delboeuf il- mindset. For instance, it is possible that the attention-
lusion, thereby affecting the perception of serving sizes. capturing tendency of highly saturated color may encour-
Extant research further indicates that emotion enhances age lower-level construal, as compared to less saturated
contrast sensitivity irrespective of attention (Phelps, Ling, color (Lee et al. 2014).
and Carrasco 2006). This suggests that the role of contrast The influence of color on perceived size may also lead
in the saturation–size effect could vary, depending on the to downstream outcomes other than the ones uncovered in
current emotional state of the viewer. Future research may the current work. For example, recent research has demon-
do more to investigate the extent to which color effects de- strated that larger sizes result in negative evaluations of
pend on the individual viewer (e.g., emotional state) or clothing (Hoegg et al. 2014). How, then, might color in-
general perceptual principles (e.g., saturated color appears form clothing size or influence perceptions of the wearer’s
to be in the foreground rather than background of a visual body? Color may have both direct and downstream influ-
scene). Either way, a role of contrast is not necessarily in- ences on consumer responses in numerous other areas as
compatible with the demonstrated role of arousal. On the well, ranging from emotion (Valdez and Mehrabian 1994)
contrary, it seems feasible that color contrast may enhance to food consumption (Van Ittersum and Wansink 2012).
arousal. Thus, the roles of contrast, arousal, attention, and Whereas emotion-related color effects can be tied to both
size perceptions may be intertwined, and more research is innate and culturally learned responses, food represents a
needed to map out the nature of these relationships. biologically and culturally vital context in which a plethora
Much like colors do not tend to exist in isolation, the in- of color effects could be explored, whether it is the food,
dividual dimensions of color do not operate in isolation. container, plates, or interior décor that displays the color in
Hue, value, and saturation all give rise to a variety of docu- question. For example, highly saturated food coloring
mented effects, and future research should explore various might decrease consumption if it causes the food to appear
combinations of levels across the color dimensions. For ex- larger, or perhaps it might increase consumption via its ef-
ample, although differential arousal stemming from differ- fect on attention.
ent hues was not documented in the current research, the Relatedly, color not only plays a vital role in object per-
effect of hue on arousal has been shown in prior work ception, but it also provides an aesthetic component that
(Bagchi and Cheema 2013; Gorn et al. 1997; Labrecque shapes how we perceive the world (Gegenfurtner and
and Milne 2012). Therefore, it also seems feasible that hue Kiper 2003). Color effects are therefore broadly applicable
could moderate the influence of saturation on perceived in contexts ranging from retail interiors to automobile exte-
size. Prior work demonstrates the effect of hue on attention riors and from packaging to screen displays. Growing re-
as well (Labrecque et al. 2013). Future research may find search areas such as design and aesthetics are emphasizing
that the strength of such influences depends on the level of the importance of color and the need to explicate its role in
value or saturation of the colors. Relatedly, in early explor- consumer behavior.
atory studies conducted by the authors, value was not held As discussed in appendix A, the importance of this topic
constant; the more saturated colors were darker than the also underscores the need for clarity in both the existing
less saturated ones. Nonetheless, despite prior findings that findings and future ones. Not only can there be uninten-
lighter-colored objects are perceived to be larger tional differences between stimuli viewed by participants
(Gundlach and Macoubrey 1931; Nakatani 1989), we due to factors such as software and screens, but there may
found that objects with darker, more highly saturated color also be individual differences in how people experience
appeared larger than objects with lighter, less saturated color. Color blindness is a salient example, but there may
color. This suggested to us that the influence of saturation be numerous types of more subtle differences as well,
may perhaps be stronger than the influence of lightness, many of which may require research to uncover.
but future research could do more to determine this ques- Additionally, the inconsistent use of color terms in the
tion. In general, such research could also help establish current literature can complicate interpretations of

Downloaded from https://academic.oup.com/jcr/article-abstract/44/2/396/2939542

by guest
on 31 March 2018
 406 JOURNAL OF CONSUMER RESEARCH

findings. Terms such as brightness, lightness, and lumi- FIGURE A1

nance are sometimes used interchangeably and other times
have different meanings (Labrecque et al. 2013; Mahnke ADDITIVE VERSUS SUBTRACTIVE COLOR SYSTEMS
1996; Semin and Palma 2014; Walker et al. 2010).
Similarly, saturation is sometimes described as chroma, pu-
rity, intensity, vibrancy, vividness, strength, or colorful-
ness, but such terms can also have separate, unique
definitions (Gorn et al. 2004; Labrecque et al. 2013;
Navalpakkam and Itti 2006). Different color models, such
as the HSL (hue-saturation-lightness) and HSV (hue-satu-
ration-value) models, also entail slightly different concep-
tualizations of saturation. As these observations illustrate,
it can be challenging to compare research findings with
such a bewildering array of terms and concepts. Scholars
in this area should therefore be especially willing to test
hypotheses even if they appear to contradict accepted wis-
dom; it is possible that the contradiction is illusory.
Researchers should also be clear about which conceptuali-
zations they are relying on, to facilitate future comparisons.
Despite more than a century of psychological color re- color component ranging from 0% to 100% in most appli-
cations. See figure A1.
search, much of this area still appears to be in its infancy.
Regardless of the technique used, color is typically
conceptualized in terms of three perceptually relevant
dimensions, but these dimensions may also vary in terms
DATA COLLECTION INFORMATION of the attributes they specify. Two common examples
are the HSV (hue-saturation-value; sometimes called
The first author collected and analyzed the data for stud- HSB: hue-saturation-brightness) and HSL (hue-satu-
ies 1 and 3–6. The second author collected and analyzed ration-lightness) models. For both models, hue ranges
the data for study 2. The data for all studies were collected from 0 degrees to 360 degrees, and saturation and value/
at Boston College. Data for studies 1 and 2 were collected lightness each range from 0% to 100% in most applica-
in fall 2015. Data for studies 3–6 were collected in fall tions. These models are both intended to be intuitive, but
2016. they emphasize different attributes. For instance, satura-
tion may be defined as a visual target’s colorfulness rela-
APPENDIX A tive to its brightness. In turn, brightness is defined as the
CONCEPTUALIZATION, appearance of light radiating from a stimulus (i.e., the
perception arising from luminance), whereas lightness is
MANIPULATION, AND MEASUREMENT defined as a stimulus’s brightness relative to the bright-
OF COLOR ness of a similarly illuminated white. Therefore,
research findings may or may not be directly compara-
Conceptual Challenges ble, depending on whether they rely on the same concep-
Color research is challenging because of conceptual confu- tualizations. See figure A2 for simplified depictions of
sion as well as difficulties in accurate manipulation and HSV and HSL models.
measurement. There exists no universally accepted stand- Adding to the confusion, the same terms are sometimes
ard for how to conceptualize color. This is partly because used with different meanings within different models. For
different conceptualizations are more applicable for differ-
ent industries and different display technologies. In com-
puter graphics, additive colors emerge through the FIGURE A2
combination of primary light colors. The RGB model (red-
green-blue) is standard in this context, with each color HSV AND HSL COLOR SPACE
component ranging from 0 to 255 in most applications.
Combining equal pairs of these three colors results in addi-
tive secondary colors—cyan, magenta, and yellow—which
are the primary subtractive colors used by professionals in
the paint or print industries. In a subtractive color system,
colors emerge when parts of the spectrum are subtracted
from white light. In the printing industry, the CMYK
(cyan-magenta-yellow-black) model is standard, with each

Downloaded from https://academic.oup.com/jcr/article-abstract/44/2/396/2939542

by guest
on 31 March 2018
 HAGTVEDT AND BRASEL 407

example, saturation is gauged differently in the HSV ver- differences in how the software specifies HSV, HSL, or
sus HSL models. In HSV, tints (i.e., pure color mixed with other measures.
white) have reduced saturation, but shades (i.e., pure color Computer-based research poses other challenges of
mixed with black) do not. In HSL, tints and shades are which researchers should be aware. For example, altera-
both fully saturated, whereas tones (i.e., mixtures with tions can occur when color stimuli are converted from
grey) are not. Relatedly, in the HSV model, low value one file type to another in electronic documents, present-
tends toward black, but high value tends toward the pure ing a potential issue both for stimuli creation and for
hue, not toward white. The Munsell color system, however, preparation of figures for article submission. Online
uses the term value much like the HSL system uses the color research often has the added complication of dif-
term lightness (with black and white at extreme ends of a ferent hardware and software used by study participants.
continuum). Similarly, the terms saturation and chroma are For instance, different screen types can give rise to var-
sometimes used interchangeably, whereas saturation iance in color display and perception. This variability
(chroma) is typically defined as the colorfulness of a stimu- comes on top of individual differences in the experience
lus relative to its own brightness (relative to the brightness of color, due to factors such as color blindness or
of a similarly illuminated white). Munsell even uses the decreased sensitivity to short-wavelength light in con-
terms interchangeably on their website while simultane- nection with age-related yellowing of the ocular lens.
ously posting a note to distinguish between the two. HSI Although random assignment should prevent systematic
(hue-saturation-intensity) represents yet another model, biases between conditions, the investigation of subtle
with intensity specifying lightness, but intensity is else- color effects may be especially susceptible to type II
where used to describe the perceptual experience of errors, so minimizing noise in the data is desirable. We
saturation. do not suggest that one cannot conduct color research
Further adding to the confusion, different authors rely on online, but one should be cognizant of the shortcomings.
different scales, which may or may not be based on the Conducting at least some of the studies in carefully con-
specific software being used. For example, the software trolled environments and with identical hardware and
used by Mehta and Zhu (2009) leads them to report specifi- software can increase the likelihood of detecting patterns
cations that do not adhere to the 0%–100% scales that are that might otherwise be missed.
typically used for saturation and lightness. Conversely, Researchers should also consider the role of other
Hagtvedt and Brasel (2016) refer to Munsell’s description attributes of color stimuli, such as their shape. In the cur-
of value as degree of lightness, but they do not use rent research, for instance, we focus on 3D objects rather
Munsell’s 0–10 scale in stimulus descriptions; their value than flat, uniform stimuli such as Munsell color chips.
measures correspond to HSV measures of 0%–100%, pre- This approach arguably has the benefit of being more
sumably because they deem the latter scale to be more relevant to everyday experience, but it also demands
intuitive. Given the use of stimuli with 100% saturation, care. Light and shade on 3D objects give rise to varying
the HSV scale for value is equivalent to the lower half of HSV values at various points on an object, so it is impor-
the Munsell scale for value, with HSV value measures cor- tant to measure the same points on each stimulus within
responding to exactly one-fifth of Munsell value measures. the same experiment. Other potential sources of inaccur-
Nonetheless, it may have been preferable to draw on one acy may be uncovered as color research progresses, so
model at a time. our final recommendation is for scholars to be vigilant
for complicating factors and to report these along with
their findings.

Recommendations
Various approaches to color research have different merits.
Several other color models exist in addition to the ones dis-
Further Reading
cussed here, and increasingly accurate and user-friendly Hall, Roy (1988), Illumination and Color in Computer
color models may emerge with time. In the meantime, the Generated Imagery, New York: Springer.
HSV and HSL models are both common and intuitive, so Joblove, George H. and Donald Greenberg (1978), “Color
for scholars who have not yet developed a preference, we Spaces for Computer Graphics,” ACM SIGGRAPH
would recommend providing one of those sets of measures. Computer Graphics, 12 (3), 20–25.
(We used HSV in the current investigation, but HSL is MacEvoy, Bruce (2016), “Color Vision,” http://www.hand
equally intuitive.) For computer-based manipulations, it print.com/LS/CVS/color.html.
could also be helpful to provide the RGB codes, as these Munsell Color, “How Color Notation Works,” http://munsell.
can be used with either model. For printed or painted stim- com/about-munsell-color/how-color-notation-works/.
uli, one might add CMYK codes. Numerous freely avail- Smith, Alvy Ray (1978), “Color Gamut Transform Pairs,”
able websites can aid with the conversion between these ACM SIGGRAPH Computer Graphics, 12 (3), 12–19.
different measures. With computer-based research, we also Stone, Maureen (2003), A Field Guide to Digital Color,
recommend specifying the software used, in case there are Natick, MA: A K Peters.

Downloaded from https://academic.oup.com/jcr/article-abstract/44/2/396/2939542

by guest
on 31 March 2018
 408 JOURNAL OF CONSUMER RESEARCH

APPENDIX B
HSV AND RGB SPECIFICATIONS FOR STIMULI

Study Hue name Saturation H S V R G B

1 Green High 117 100 70 9 179 0

Low 117 50 70 94 179 89
2 Blue High 200 100 63 0 107 161
Low 200 50 63 80 134 161
Red High 0 100 54 138 0 0
Low 0 50 54 138 69 69
3 Red High 356 100 70 179 0 12
Low 356 50 70 179 89 95
4 Orange High 22 100 80 204 75 0
Low 22 50 80 204 139 102
5 Orange Highh 27 100 80 204 92 0
Lowl 27 50 80 204 148 102
6 Violet High 280 100 75 127 0 191
Low 280 50 75 159 96 191
Note.—Conversions to CMYK measures:
h
(0, .55, 1, .2),
l
(0, .275, .5, .2).

Downloaded from https://academic.oup.com/jcr/article-abstract/44/2/396/2939542

by guest
on 31 March 2018
 HAGTVEDT AND BRASEL 409

APPENDIX C
STIMULI FOR STUDIES 1–4

Study 1: High saturation Study 1: Low saturation

Study 3: High saturation

Study 2: Sample eye-tracking stimulus

Study 3: Low saturation

Study 4: High saturation

Downloaded from https://academic.oup.com/jcr/article-abstract/44/2/396/2939542

by guest
on 31 March 2018
 410 JOURNAL OF CONSUMER RESEARCH

Study 4: Low saturation

APPENDIX D
STIMULI FOR STUDIES 5 AND 6

Study 5: High versus low saturation (image adjusted to compensate for lighting)

Downloaded from https://academic.oup.com/jcr/article-abstract/44/2/396/2939542

by guest
on 31 March 2018
 HAGTVEDT AND BRASEL 411

as the dependent variable, and attention as the mediator.

The indirect effect was significant (B ¼ .48, SE ¼ .22;
95% CI ¼ .07, .95), indicating mediation, whereas the
direct effect became nonsignificant (B ¼ .44, SE ¼ .27;
95% CI ¼ –.10, .98).
Finally, to test the full mediational pathway, a similar
bootstrap estimation (PROCESS model 6; Hayes 2013)
was conducted with saturation as the independent variable,
size estimate as the dependent variable, and attention and
arousal as successive mediators. The total indirect effect
was significant (B ¼ .53, SE ¼ .22; 95% CI ¼ .16, 1.02),
indicating mediation, whereas the direct effect became
nonsignificant (B ¼ .34, SE ¼ .28; 95% CI ¼ –.22, .90).
More specifically, the indirect effect via both mediators
was significant (B ¼ .18, SE ¼ .10; 95% CI ¼ .03, .45), as
was the indirect effect via attention alone (B ¼ .18, SE ¼
.12; 95% CI ¼ .01, .52), whereas the indirect effect via
Study 6: High saturation arousal alone became nonsignificant (B ¼ .17, SE ¼ .12;
95% CI ¼ –.02, .47).
These results mirror those based on the hypothesized
mediational pathway. In other words, the account in which
arousal precedes attention and the account in which atten-
tion precedes arousal are equally plausible; this question of
order remains for future research.

REFERENCES
Adolphs, Ralph and Antonio R. Damasio (2001), “The Interaction
of Affect and Cognition: A Neurobiological Perspective,” in
Handbook of Affect and Social Cognition, ed. Joseph P.
Forgas, Mahwah, NJ: Erlbaum Associations, 27–49.
_
Aydinoglu, NIlüfer Z. and Aradhna Krishna (2011), “Guiltless
Gluttony: The Asymmetric Effect of Size Labels on Size
Perceptions and Consumption,” Journal of Consumer
Research, 37 (6), 1095–112.
Aziz, M. Zaheer, B€arbel Mertsching, M. Salah E-N Shafik, and
Ralf Stemmer (2006), “Evaluation of Visual Attention
Models for Robots,” Proceedings of the Fourth IEEE
Study 6: Low saturation International Conference on Computer Vision Systems,
Washington, DC: IEEE Computer Society.
Bagchi, Rajesh and Amar Cheema (2012), “The Effect of Red
Background Color on Willingness-to-Pay: The Moderating
APPENDIX E Role of Selling Mechanism,” Journal of Consumer Research,
ALTERNATIVE MEDIATION 39 (5), 947–60.
FOR STUDY 3 Beatty, Jackson and Brennis Lucero-Wagoner (2000), “The
Pupillary System,” in Handbook of Psychophysiology, ed.
To test whether arousal mediated the effect of saturation John T. Cacioppo, Louis G. Tassinary, and Gary G. Berntson,
on size estimates, bootstrap estimation (PROCESS model Hillsdale, NJ: Cambridge University Press, 142–62.
4; Hayes 2013) with 10,000 resamples was conducted with Bradley, Margaret M., Laura Miccoli, Miguel A. Escrig, and Peter
saturation as the independent variable, size estimate as the J. Lang (2008), “The Pupil as a Measure of Emotional
dependent variable, and arousal as the mediator. The indi- Arousal and Autonomic Activation,” Psychophysiology, 45
(4), 602–7.
rect effect was significant (B ¼ .49, SE ¼ .21; 95% CI ¼ Brasel, S. Adam and Henrik Hagtvedt (2016), “Living Brands:
.15, .96), indicating mediation, whereas the direct Consumer Responses to Animated Brand Logos,” Journal of
effect became nonsignificant (B ¼ .38, SE ¼ .29; 95% the Academy of Marketing Science, 44 (5), 639–53.
CI ¼ –.19, .95). Braun, Jochen (1994), “Visual Search among Items of Different
To test whether attention mediated the effect of satura- Salience: Removal of Visual Attention Mimics a Lesion in
tion on arousal, a similar bootstrap estimation was con- Extrastriate Area V4,” Journal of Neuroscience, 14 (2),
ducted with saturation as the independent variable, arousal 554–67.

Downloaded from https://academic.oup.com/jcr/article-abstract/44/2/396/2939542

by guest
on 31 March 2018
 412 JOURNAL OF CONSUMER RESEARCH

Camgöz, Nilgün, Cengiz Yener, and Dilek Güvenç (2004), Expected Clothing Sizes,” Journal of Consumer Psychology,
“Effects of Hue, Saturation, and Brightness: Part 2: 24 (1), 70–8.
Attention,” Color Research & Application, 29 (1) 20–8. Kareklas, Ioannis, Frédéric F. Brunel, and Robin A. Coulter
De Bock, Tine, Mario Pandelaere, and Patrick Van Kenhove (2014), “Judgment Is Not Color Blind: The Impact of
(2013), “When Colors Backfire: The Impact of Color Cues on Automatic Color Preference on Product and Advertising
Moral Judgment,” Journal of Consumer Psychology, 23 (3), Preferences,” Journal of Consumer Psychology, 24 (1),
341–8. 87–95.
Deng, Xiaoyan, Sam K. Hui, and J. Wesley Hutchinson (2010), Kupor, Daniella M. and Zakary L. Tormala (2015), “Persuasion,
“Consumer Preferences for Color Combinations: An Interrupted: The Effect of Momentary Interruptions on
Empirical Analysis of Similarity-Based Color Message Processing and Persuasion,” Journal of Consumer
Relationships,” Journal of Consumer Psychology, 20 (4), Research, 42 (2), 300–15.
476–84. Labrecque, Lauren I. and George R. Milne (2012), “Exciting Red
Dresp-Langley, Birgitta, and Adam Reeves (2014), “Effects of and Competent Blue: The Importance of Color in
Saturation and Contrast Polarity on the Figure-Ground Marketing,” Journal of the Academy of Marketing Science,
Organization of Color on Gray,” Frontiers in Psychology, 5 40 (5), 711–27.
(October). Labrecque, Lauren I., Vanessa M. Patrick, and George R. Milne
Egusa, Hiroyuki (1983), “Effects of Brightness, Hue, and (2013), “The Marketer’s Prismatic Palette: A Review of
Saturation on Perceived Depth between Adjacent Regions in Color Research and Future Directions,” Psychology and
the Visual Field,” Perception, 12 (2), 167–75. Marketing, 30 (2), 187–202.
Elliot, Andrew J. and Markus A. Maier (2014), “Color Lee, Hyojin, Xiaoyan Deng, H. Rao Unnava, and Kentaro Fujita
Psychology: Effects of Perceiving Color on Psychological (2014), “Monochrome Forests and Colorful Trees: The Effect
Functioning in Humans,” Annual Review of Psychology, 65 of Black-and-White versus Color Imagery on Construal
(January), 95–120. Level,” Journal of Consumer Research, 41 (4), 1015–32.
Folkes, Valerie and Sashi Matta (2004), “The Effect of Package MacLachlan, Colleen and Howard C. Howland (2002), “Normal
Shape on Consumers’ Judgments of Product Volume: Values and Standard Deviations for Pupil Diameter and
Attention as a Mental Contaminant,” Journal of Consumer Interpupillary Distance in Subjects Aged 1 Month to 19
Research, 31 (2), 390–401. Years,” Opthalmic & Physiological Optics, 22 (3), 175–82.
Frey, Hans-Peter, Christian Honey, and Peter König (2008), Mahnke, Frank H. (1996), Color, Environment, and Human
“What’s Color Got to Do with It? The Influence of Color on Response, New York: John Wiley & Sons.
Visual Attention in Different Categories,” Journal of Vision, McConnell, Meghan M. and David I. Shore (2011), “Upbeat and
8 (14), 1–17. Happy: Arousal as an Important Factor in Studying
Fuller, Stuart and Marisa Carrrasco (2006), “Exogenous Attention Attention,” Cognition and Emotion, 25 (7), 1184–95.
and Color Perception: Performance and Appearance of Mehta, Ravi and Rui (Juliet) Zhu (2009), “Blue or Red? Exploring
Saturation and Hue,” Vision Research, 46 (23), 4032–47. the Effect of Color on Cognitive Task Performances,”
Gable, Philip A. and Eddie Harmon-Jones (2008), “Approach- Science, 323 (5918), 1226–9.
Motivated Positive Affect Reduces Breadth of Attention,” Meyers-Levy, Joan and Laura A. Peracchio (1995), “Understanding
Psychological Science, 19 (5), 476–82. the Effects of Color: How the Correspondence between
Gegenfurtner, Karl R. and Daniel C. Kiper (2003), “Color Available and Required Resources Affects Attitudes,” Journal
Vision,” Annual Review of Neuroscience, 26 (1), 181–206. of Consumer Research, 22 (2), 121–38.
Gobell, Joetta and Marisa Carrasco (2005), “Attention Alters the Michael, George A. and German Galvez-Garcıa (2011),
Appearance of Spatial Frequency and Gap Size,” “Salience-Based Progression of Visual Attention,”
Psychological Science, 16 (8), 644–51. Behavioural Brain Research, 224 (1), 87–99.
Gorn, Gerald J., Amitava Chattopadhyay, Jaideep Sengupta, and Mizzi, Rapha€el and George A. Michael (2014), “The Role of the
Shashank Tripathi (2004), “Waiting for the Web: How Collicular Pathway in the Salience-Based Progression of
Screen Color Affects Time Perception,” Journal of Visual Attention,” Behavioural Brain Research, 270
Marketing Research, 41 (2), 215–25. (August), 330–8.
Gorn, Gerald J., Amitava Chattopadhyay, Tracey Yi, and Darren Nakatani, Katsuya (1989), “Fixed Set in the Perception of Size in
W. Dahl (1997), “Effects of Color as an Executional Cue in Relation to Lightness,” Perceptual and Motor Skills, 68 (2),
Advertising: They’re in the Shade,” Management Science, 43 415–22.
(10), 1387–400. Navalpakkam, Vidhya and Laurent Itti (2006), “Top-Down
Gundlach, Carol and Constance Macoubrey (1931), “The Effect of Attention Selection Is Fine Grained,” Journal of Vision, 6
Color on Apparent Size,” American Journal of Psychology, (11), 1180–93.
43 (1), 109–11. €
Ohman, Arne, Anders Flykt, and Francisco Esteves (2001),
Hagtvedt, Henrik and S. Adam Brasel (2016), “Cross-Modal “Emotion Drives Attention: Detecting the Snake in the
Communication: Sound Frequency Influences Consumer Grass,” Journal of Experimental Psychology: General, 130
Responses to Color Lightness,” Journal of Marketing (3), 466–78.
Research, 53 (4), 551–62. Ordabayeva, Nailya and Pierre Chandon (2013), “Predicting and
Hayes, Andrew F. (2013), Introduction to Mediation, Moderation, Managing Consumers’ Package Size Impressions,” Journal
and Conditional Process Analysis: A Regression-Based of Marketing, 77 (5), 123–37.
Approach, New York: Guilford. Phelps, Elizabeth A., Sam Ling, and Marisa Carrasco (2006),
Hoegg, JoAndrea, Maura L. Scott, Andrea C. Morales, and Darren “Emotion Facilitates Perception and Potentiates the
W. Dahl (2014), “The Flip Side of Vanity Sizing: How Perceptual Benefits of Attention,” Psychological Science, 17
Consumers Respond to and Compensate for Larger than (4), 292–9.

Downloaded from https://academic.oup.com/jcr/article-abstract/44/2/396/2939542

by guest
on 31 March 2018
 HAGTVEDT AND BRASEL 413

Pieters, Rik and Michel Wedel (2004), “Attention Capture and Tedford, W. H., S. L. Berquist, and W. E. Flynn (1977), “The
Transfer in Advertising: Brand, Pictorial, and Text-Size Size-Color Illusion,” Journal of General Psychology, 97 (1),
Effects,” Journal of Marketing, 68 (2), 36–50. 145–9.
Puccinelli, Nancy M., Rajesh Chandrashekaran, Dhruv Grewal, Valdez, Patricia and Albert Mehrabian (1994), “Effects of Color
and Rajneesh Suri (2013), “Are Men Seduced by Red? The on Emotions,” Journal of Experimental Psychology:
Effect of Red Versus Black Prices on Price Perceptions,” General, 123 (4), 394–409.
Journal of Retailing, 89 (2), 115–25. Van den Bergh, Bram, Julien Schmitt, and Luk Warlop (2011),
Russell, James A., Anna Weiss, and Gerald A. Mendelsohn (1989), “Embodied Myopia,” Journal of Marketing Research, 48 (6),
“Affect Grid: A Single-Item Scale of Pleasure and Arousal,” 1033–44.
Journal of Personality and Social Psychology, 57 (3), 493–502. Van Ittersum, Koert and Brian Wansink (2012), “Plate Size and
Semin, Gun R. and Thomas A. Palma (2014), “Why the Bride Color Suggestibility: The Delboeuf Illusion’s Bias on
Wears White: Grounding Gender with Whiteness,” Journal Serving and Eating Behavior,” Journal of Consumer
of Consumer Psychology, 24 (2), 217–25. Research, 39 (2), 215–28.
Sevilla, Julio and Barbara E. Kahn (2014), “The Completeness Walker, Peter, Brian J. Francis, and Leanne Walker (2010),
Heuristic: Product Shape Completeness Influences Size “The Brightness-Weight Illusion: Darker Objects Look
Perceptions, Preference, and Consumption,” Journal of Heavier but Feel Lighter,” Experimental Psychology, 57
Marketing Research, 51 (1), 57–68. (6), 462–9.
Spence, Charles (2011), “Crossmodal Correspondences: A Tutorial Wansink, Brian (2006), Mindless Eating: Why We Eat More than
Review,” Attention, Perception, & Psychophysics, 73 (4), 971–95. We Think, New York: Bantam Books.

Downloaded from https://academic.oup.com/jcr/article-abstract/44/2/396/2939542

by guest
on 31 March 2018