General Methods

Do color preferences vary for di�erent objects? Color preferences vary across objects

“Imagined” object preferences generalize well What determines object color preferences?

References and Acknowledgments

Functional reasons for object color preferences

Object Color PreferencesKaren B. Schloss1, Eli D. Strauss2, and Stephen E. Palmer1,2

1Department of Psychology, 2Program in Cognitive Science, UC Berkeley

R O Y H G C B P-50

-25

0

25

50

Hue

Aver

age

Pref

eren

ce R

atin

g SquareT-ShirtTie/ScarfShirt/Blouse

WallTrimCouchPillow

Dark Medium Light-60

-30

0

30

Aver

age

Pref

eren

ce R

atin

g

Lightness LevelDark Medium Light

Lightness LevelDark Medium Light

Lightness Level

SquareT-Shirt

Tie/Scarf

Shirt/Blouse

WallTrim

Couch

PillowWallTrim

−.50 −.25 .00 .25 .50 .75−.50

−.25

.00

.25

.50

Square

Wall

Trim

CouchPillow

T-shirt

Blouse/ShirtTie/Scarf

Dimension 1SaturatedDesaturated

Ligh

tD

ark

Dim

ensi

on 2

R O Y H G C B P-30

0

30

60

Hue

Pref

eren

ceD

iffer

ence

R O Y H G C B P-30

0

30

60

HueR O Y H G C B P

-30

0

30

HueR O Y H G C B P

-30

0

30

60

HueR O Y H G C B P

-30

0

30

60

Hue

VW Bugr = .89

T- Shirtr = .97

Couchr = .96

Wallr = .94

Sedanr = .98

60

0

20

40

60

80

100

VW Bug T-shirt Wall Sedan Couch

Preference

Open/Spacious

Appropriate Appropriate Appropriate

Appropriate

Appropriate

LuxuriousRelaxing

Preference

PreferencePreference

Perc

ent E

xpla

ined

SaturatedLightMutedDark

R O Y H G C B P-50

-25

0

25

50

75

Hue

Mea

n C

olor

Pre

fere

nce

Hue. Peak at blue, trough around yellow - chartreuse.

Saturation. Saturated colors preferred to Light and Muted colors.

Lightness. Preference for dark red and dark green versus dislike for dark yellow (olive) and dark orange (brown) relative to Light and Muted colors. (Palmer & Schloss, 2010)

Research Questions:1. Do abstract color preferences generalize to other objects/contexts?

2. Why might color preferences vary across object contexts? Appropriateness/conventionality (Sivik, 1974; Whit�eld & Slatter, 1978; Taft, 1997)

Desired emotional experience (Manav, 2007; Destefani and Whit�eld, 2008)

3. How good are people at imagining their color preference for a particular object without seeing the colored object?

US preferences for abstract “contextless” colored squares

Not for objects with diagnostic colors ...

... but what about objects that can be any color?

SATURATED

MUTED

LIGHT

DARK

4 unique hues: red yellow green blue

4 angle bisectors: orange chartreuse cyan purple

4 saturation/lightness levels (”cuts”): saturated light muted dark

5 achromatic colors

Consistent Preferences for Hue

Multidimensional scaling (MDS) based on color preference correlations for each pair of objects

Variable Preferences for Saturation and Lightness

(big red objects)

(purple objects)

Dimension 1 correlates (r = .96) with the di�erence in preference between saturated and muted colors for each object.

Dimension 2 correlates (r = .94) with the slope of preference functions over lightness levels .

Note: These dimensions are de�ned by color preferences for the objects, not by inherent color properties of the objects

stress = .008

Very Much(+100)

Not at All(-100)

How much do you like this color for ...

Very Much(+100)

Not at All(-100)

ImaginedExp 1 & 3

Very Much(+100)

Not at All(-100)

ImaginedExp 2

DepictedExp 2 & 3

PhysicalExp 3

...T-shirts?

...walls?

...trim?

...couches?

...throw pillows?

...dress shirts/blouses?

...ties/scarfs?

...luxury sedans?

...VW Bugs?

Imagined and depicted object preferences are highly correlated for all objects, but there are some reliable di�erences in preference for pictures of colored objects.

However, when there is a discrepancy between imagined and depicted color preferences, it is not clear which judgment better re�ects color preferences for physical objects.

Experiment 1: Imagined Object Color Preferences

Experiment 2: Imagined vs. Depicted Object Color Preferences

Experiment 3: Imagined vs. Depicted vs. Physical T-shirt Color Preferences

Depicted dark (p < .001) and red (p<.006) walls more preferred

Depicted Bugs in the orange to green range more preferred

Depicted Sedans preferred (p<.001)

Depicted red (p<.006) couches more preferred

No di�erences

T-shirt color preferences are consistent across tasks imagined vs. depicted: r = .97 imagined vs. physical: r = .94 depicted vs. physical: r = .95

Participants reported the following features as important when choosing object colors: 1. Appropriate (all) 4. Open/Spacious (walls)2. Relaxing (all) 5. Dirt Hiding (couches)3. Luxurious (cars) 6. Not Police Attracting (cars)

Other participants rated each color for each of these six features, for appropriate objects.

For a wall color, how......appropriate?...open/spacious?...etc.is this color?

Functional features explained variance in addition to contextless preferences:

The importance of contextless preference is inversely related to importance of appropriateness, consistent with Taft (1997) and Sivik (1974).

Very Much(+100)

Not at All(-100)

R O Y H G C B P-90-60-30

03060

Hue

Wall

Appr

opria

tene

ss

R O Y H G C B PHue

Couch

R O Y H G C B PHue

Sedan

R O Y H G C B PHue

VW Bug

R O Y H G C B PHue

T-shirtSLMD

Why?

1. They may re�ect the ecological statistics of the colors of that type of object (Bayesian priors).

2. Saturated colors may seem inappropriate because they are “loud” and “�ashy,” which can be undesirable.

3. Objects in appropriate colors may be more recogniz-able and thus more �uently processed (Reber et al. 2004).

Appropriateness varies across objects:

Low-level biological factors: e.g., cone contrast theory? (Hurlbert & Ling, 2007)

Unlikely, because cone responses would not depend on the object viewed.

Ecological factors: e.g., ecological valence theory (EVT)? (Palmer & Schloss, 2010)?

Lightness Invariant Darker Preferred Lighter Preferred

Depicted

Imagined

Highly likely, because EVT is based on the degree of liking/disliking colored objects. However, a direct application is inappropriate because people do experience every object in every color.

The EVT must be augmented to include e�ects of desired experiences with objects.

desired experience functional perceptual social

object preferences

contextless colorpreferences

object color preferences

The presented research is currently in press: Schloss, K. B., Strauss, E. D., & Palmer, S. E. (in press). Object color preferences. Color Research & Application.

References:de Destefani, L. R. G. & Whit�eld, T. W. A. (2008). Esthetic decision-making: How do people select colours for real settings? Color Research & Application, 33, 55-60.Hurlbert, A. C., Ling, Y. (2007). Biological components of sex di�erences in color preference, Current Biology, 17, 623-625.Manav, B. (2007). Color-emotion associations and color preferences: A case study for residences. Color Research & Application, 32, 144-150.Palmer, S. E., Schloss, K. B. (2010). An ecological valence theory of human color preference. Proceedings of the National Academy of Sciences, 107, 8877-8882.Reber, R., Schwarz, N., & Winkielman, P. (2004). Processing �uency and aesthetic pleasure: Is beauty in the perceiver's processing experience? Personality and Social Psychology Review, 8, 364-382.Sivik, L. (1976). Colour meaning and perceptual colour dimensions: A study of exterior colours, Göteborg Psychological Reports, 4, No. 11.Taft, C. (1997). Color meaning and context: Comparisons of semantic ratings of colors on samples and objects. Color Research & Application, 22, 40-50.Whit�eld, T. W. A. & Slatter, P. (1978). The evaluation of architectural interior colour as a function of style of furnishings: Categorization e�ects. Scandinavian Journal of Psychology, 19, 51-255.

We thank Jonathan Sammartino, Matthew Barker-Ben�eld, Rosa Poggesi, Joseph Austerweil, Thomas Langlois, Will Griscom, Saki Wang, Ruth Ezra, Madison Zeller, Mathilde Heinemann, and Jackson Jewett for their help with this research. This project was supported by NSF Grant 0745820 and 1059088 and a Google Gift to SEP.