BRIEF REPORT

Joint Effects of Emotion and Color on Memory

Christof Kuhbandner and Reinhard PekrunUniversity of Munich

Numerous studies have shown that memory is enhanced for emotionally negative and positive informa-tion relative to neutral information. We examined whether emotion-induced memory enhancement isinfluenced by low-level perceptual attributes such as color. Because in everyday life red is often used asa warning signal, whereas green signals security, we hypothesized that red might enhance memory fornegative information and green memory for positive information. To capture the signaling function ofcolors, we measured memory for words standing out from the context by color, and manipulated the colorand emotional significance of the outstanding words. Making words outstanding by color stronglyenhanced memory, replicating the well-known von Restorff effect. Furthermore, memory for coloredwords was further increased by emotional significance, replicating the memory-enhancing effect ofemotion. Most intriguingly, the effects of emotion on memory additionally depended on color type. Redstrongly increased memory for negative words, whereas green strongly increased memory for positivewords. These findings provide the first evidence that emotion-induced memory enhancement is influ-enced by color and demonstrate that different colors can have different functions in human memory.

Keywords: emotion, color, memory, von Restorff effect, isolation effect

One prominent factor influencing the likelihood of rememberinginformation is emotional salience. Numerous studies have shownthat recall rates are higher for emotionally negative and positivestimuli compared with neutral stimuli. The findings were consis-tent across a variety of stimuli types, including words, sentences,and pictures (see, e.g., Hamann, 2001, and Reisberg & Heuer,2004, for reviews). Given the robustness of emotion-inducedmemory enhancement, an interesting question arises: Are there anylow-level stimulus features that can influence the effect of emo-tional salience on memory? An answer to this question would alsobe of relevance for applied fields in which it is important toeffectively visualize information.

One low-level stimulus property that might influence memoryfor emotional information is color. In everyday life, colors are notused arbitrarily. Instead, there are regularities in the usage ofcolors to convey information about situational requirements, suchas the usage of red and green to signal stop and go. The colorsred and green seem also to have relevance in emotional contexts.Typically, red is used as a signal for negative, especially threat-

ening, information (e.g., alarms, warning signals), whereas greensignals security and safety (Travis, 1991). Given the ubiquity ofsuch color usage in everyday life, red and green might haveacquired the function of implicit cues alerting the perceiver toimminent danger or potential benefits, which might, in turn, in-tensify the impact of negative, respectively positive, information(e.g., Bower, 1981). Indeed, first evidence in the domain of stim-ulus categorization suggests that emotional connotations of red andgreen can differentially influence emotional processing. Valencejudgments seem to be speeded up when negative words are pre-sented in red and positive words in green (Moller, Elliot, & Maier,2009). Accordingly, it might be that red and green differentiallyenhance memory for negative and positive stimuli.

The aim of the present study was to examine color influences onemotion-induced memory enhancement. In order to establish asituation in which colors are experienced as important signals, wepresented lists of black-colored words to participants that con-tained one critical word that stood out from the context by color.To examine the effects of different colors on memory for emo-tional information, the critical word was either colored red, green,or (as a control color) blue, and either neutral, negative, or positive(see Figure 1A). It is well-known that memory for perceptuallyoutstanding information is generally enhanced, a phenomenoncalled the von Restorff or isolation effect (von Restorff, 1933; seeHunt, 1995, for a review). Surprisingly, although colors have oftenbeen used in previous research to make stimuli outstanding, to ourknowledge neither the memory effects of different colors nor thememory effects of emotional significance of perceptually out-standing stimuli have been addressed in previous research.

We expected that making words outstanding by color wouldgenerally enhance memory for these words, replicating the typical

This article was published Online First March 25, 2013.Christof Kuhbandner and Reinhard Pekrun, Department of Psychology,

University of Munich, Munich, Germany.This research was supported by a grant entitled Developmental and

Educational Neurosciences awarded to B. Sodian and R. Pekrun from theprogram LMUexcellent within the framework of the German ExcellenceInitiative. We thank Alexandra Mller for collecting the data.

Correspondence concerning this article should be addressed to ChristofKuhbandner, Department of Psychology, University of Munich, Leopold-strasse 13, 80802 Munich, Germany. E-mail: christof.kuhbandner@psy.lmu.de

This

docu

men

tis

copy

right

edby

the

Am

eric

anPs

ycho

logi

calA

ssoc

iatio

nor

one

ofi

tsal

lied

publ

isher

s.Th

isar

ticle

isin

tend

edso

lely

fort

hepe

rson

aluse

oft

hein

divi

dual

use

ran

dis

not

tobe

diss

emin

ated

broa

dly.

Emotion 2013 American Psychological Association2013, Vol. 13, No. 3, 375379 1528-3542/13/$12.00 DOI: 10.1037/a0031821

375

von Restorff effect. Moreover, we expected that emotional signif-icance of the perceptually outstanding words would further in-crease the von Restorff effect. Such a combined effect of color andemotional significance would demonstrate that emotional saliencecan enhance memory beyond the memory-enhancing effect ofperceptual salience. However, if red and green differentially influ-ence memory for negative and positive information, then thereshould be an interaction between the effects of color and emotionalsalience. Negative words should be remembered best when coloredred, whereas positive words should be remembered best whencolored green.

Method

Participants

Forty-eight undergraduate students (42 females, mean age 22.6 years, SD 4.2) at a German university participated forcourse credit. All participants were tested individually and werenot redgreen colorblind.

Materials and ProcedureParticipants were asked to study and recall 30 successively

presented lists of words. Twenty-seven of the lists contained 11black-colored words and one red-, green-, or blue-colored word.

The colored word of a list was randomly presented at the 5th, 6th,7th, or 8th serial position, and the mean memory performance forblack words across these positions was used as a baseline for thecolored words. The remaining three lists contained only blackwords to include an additional control condition. Memory perfor-mance for black words presented at the 5th, 6th, 7th, or 8th serialposition did not differ between lists containing one colored wordand lists containing only black words, t(47) 0.89, p .376.All words were five- to six-letter German nouns. The black wordswere randomly drawn from a pool of 333 neutral words (takenfrom the CELEX database using WordGen v1.0 software toolbox;Duyck, Desmet, Verbeke, & Brysbaert, 2004). The colored wordswere randomly drawn from a pool of 27 words that differed inemotional valence (see the Appendix). Nine of the colored wordswere negative, nine were positive, and nine were neutral. The threetypes of colored words were matched on word length and wordfrequency (Google Ngram Viewer; Michel et al., 2011).

Three of the words of an emotional type were presented in red,three in green, and three in blue. The assignment of colors to thesewords was counterbalanced across participants. The colorimetricproperties of the three colors were chosen using a spectrometer(i1Pro, X-rite Inc., Grandville, Michigan) according to the LChcolor model. In order to account for possible confounding effectsof lightness and chroma, red, green, and blue hues were selectedthat were equal on the parameters L and C (red: LCh[47.0, 63.1,

Figure 1. Example of one experimental trial. (A) Participants were presented sequences of 12 words containing11 neutral black words and 1 red, green, or blue colored word that was either emotionally neutral, negative, orpositive. (B) Memory performance as a function of serial position and color of the outstanding word for neutral(left panel), negative (middle panel), and positive words (right panel). Note that the outstanding word wasactually randomly presented at serial position 5, 6, 7, or 8 (shaded in gray); the black baselines show memoryperformance for black neutral words presented at the same serial positions.

This

docu

men

tis

copy

right

edby

the

Am

eric

anPs

ycho

logi

calA

ssoc

iatio

nor

one

ofi

tsal

lied

publ

isher

s.Th

isar

ticle

isin

tend

edso

lely

fort

hepe

rson

aluse

oft

hein

divi

dual

use

ran

dis

not

tobe

diss

emin

ated

broa

dly.

376 KUHBANDNER AND PEKRUN

14.9], green: LCh[47.2, 61.9, 152.2], blue: LCh[47.3, 61.8,263.1]). In this way, colors were chosen that were equally percep-tually different from black (red: E 78.7; green: E 77.9;blue: E 77.7).

All study words were presented for 1.5 s without an interstimu-lus interval. Immediately after studying each list, participants wereasked to recall the words of the list in any order for 45 s. The orderof list presentation was randomized. After studying and recalling10 lists, there always was a short break of 1 min. After theexperiment, in order to assess the emotional valence of the words,participants were presented the colored words (in black color),again with the instruction to rate their valence on a scale rangingfrom 1 (very unpleasant) to 5 (very pleasant). To take into accountthe possibility that the prior study and recall of the words mighthave affected participants ratings of valence, and to determinewhether negative and positive words also differed in arousal, weadditionally asked eight independent judges (of the same agegroup and from the same population as the experimental partici-pants) to rate each word on valence (1 extremely unpleasant to7 extremely pleasant) and arousal (1 low arousal to 7 higharousal).

Results

Emotional Ratings

As indicated by the ratings of the participants, the three emo-tional types of colored items varied significantly in their emotionalratings. Compared with neutral words (M 3.17), negative wordswere rated much more unpleasant (M 1.31), t(47) 32.18, p .001, d 4.85,1 and positive words much more pleasant (M 4.66), t(47) 33.27, p .001, d 5.03. This difference invalence was confirmed by the ratings of the independent judges.Compared with neutral words (M 3.91), negative words wererated much more unpleasant (M 1.32), t(7) 25.64, p .001,d 9.38, and positive words much more pleasant (M 6.44),t(7) 16.79, p .001, d 6.00. The arousal ratings of theindependent judges revealed that both negative words (M 5.68)and positive words (M 6.03) were rated as more arousing thanneutral words (M 2.72), t(7) 6.84, p .001, d 2.45, andt(7) 9.67, p .001, d 3.92, respectively. Arousal did notdiffer between negative and positive words, t(7) 1.52, p .172, d 0.60.

Effects of Color and Emotional ValenceMemory performance as a function of serial position, color of

the critical word, and emotional valence is shown in Figure 1B.Memory for words standing out by color was generally muchhigher compared with black words presented at the same serialpositions, all ts 4.80, all ps .001. To examine the effects ofemotion and color type on memory for words standing out bycolor, an analysis of variance with the factors color type (red,green, blue) and emotional valence (negative, positive, neutral)was conducted. The analysis revealed significant main effects ofcolor type, F(2, 94) 4.51, p .013, p2 .09, and emotionalvalence, F(2, 94) 24.93, p .001, p2 .35. Both red and greenwords were better remembered than blue words, t(47) 2.77, p .008, d 0.40, and t(47) 2.29, p .027, d 0.34, respectively,

and both negative and positive words were better remembered thanneutral words, t(47) 5.84, p .001, d 0.84, and t(47) 5.54,p .001, d 0.80, respectively. However, these main effectswere qualified by a significant Color Type Emotional Valenceinteraction, F(4, 188) 5.38, p .001, p2 .10. Whereasmemory for neutral words was not influenced by color, F(2, 94)1.41, p .25, memory for negative words was boosted by redcolor, compared with green color, t(47) 4.08, p .001, d 0.59, and blue color, t(47) 3.16, p .003, d 0.46. By contrast,memory for positive words was boosted by green color, comparedwith red color, t(47) 2.68, p .010, d 0.39, and blue color,t(47) 3.44, p .001, d 0.50.

DiscussionThe present findings demonstrate that emotion-induced memory

enhancement can be influenced by color. We employed a typicalvon Restorff paradigm in which participants memory for wordsstanding out from the context by color was examined. Replicatingthe von Restorff effect, we found that memory for colored wordswas strongly enhanced. Furthermore, in line with previous findingsdocumenting emotion-induced memory enhancement, our resultsshowed that memory for colored words was further increased byemotional significance. Both negative and positive words werebetter remembered than neutral words. Finally, most intriguingly,our findings demonstrate that the memory-enhancing effect ofemotional significance is differentially influenced by differentcolors. Red strongly enhanced memory for negative words,whereas green strongly enhanced memory for positive words.

This pattern of results mirrors the conventional use of red andgreen in everyday life to signal that a situation is potentiallythreatening or beneficial (Travis, 1991). Accordingly, based onsocietal learning, red and green might function as implicit cuesalerting the perceiver to potential dangers or benefits, which might,in turn, intensify the impact of negative, respectively positive,information (Bower, 1981). Indeed, recent findings demonstratethat viewing red can have effects that are typically observed inthreatening situations, such as a decreased heart rate variability(Elliot, Payen, Brisswalter, Cury, & Thayer, 2011) or an enhancedforce and velocity of motor output (Elliot & Aarts, 2011). As mostof the recent studies on the effects of red on psychological func-tioning have used green only sometimes as a control color, con-clusive evidence concerning the effects of green is lacking. How-ever, a few studies on physiological reactions to different colorssuggest that green is associated with pleasant, low-arousal affec-tive states (e.g., Jacobs & Hustmyer, 1974; Wilson, 1966), al-though these studiesin contrast to recent studies on the effects ofreddid not control for confounding effects of lightness orchroma.

The idea that colors having specific emotional connotations caninfluence emotional processing is also supported by previous stud-ies on emotional valence judgments. For instance, it has beenshown that valence judgments are speeded up when negativewords are presented in black, and positive words in white, a patternreflecting the tendency of people to view dark colors as negativeand light colors as positive (Meier, Robinson, & Clore, 2004;

1 To account for repeated measures, all reported effect sizes were cal-culated using Morris and DeShons (2002) Equation 8.

This

docu

men

tis

copy

right

edby

the

Am

eric

anPs

ycho

logi

calA

ssoc

iatio

nor

one

ofi

tsal

lied

publ

isher

s.Th

isar

ticle

isin

tend

edso

lely

fort

hepe

rson

aluse

oft

hein

divi

dual

use

ran

dis

not

tobe

diss

emin

ated

broa

dly.

377COLOR AND EMOTIONAL MEMORY

Okubo & Ishikawa, 2011). Moreover, findings that were generallyconsistent with the present results on memory effects were foundwhen manipulating hue instead of lightness. Compared with awhite baseline, red facilitated the evaluation of negative words andgreen the evaluation of positive words (Moller et al., 2009).However, the effect of green seemed to be restricted to success-related positive words, and red seemed not only to facilitatevalence judgments for negative words but also to impair valencejudgments for positive words (compared with the white baseline).The findings for positive words are difficult to interpret, however.Moller and colleagues used white as a baseline color, which hasbeen shown to facilitate evaluative judgments of positive words aswell (Meier et al., 2004; Okubo & Ishikawa, 2011). Accordingly,the use of a positively biased color baseline may have maskedstronger effects of green.

One noteworthy aspect of the present results is that emotion-induced memory enhancement was found, although both emo-tional and neutral von Restorff words were highly distinct from theother words in terms of perceptual salience. This finding demon-strates that emotional salience can enhance memory over andabove the memory-enhancing effect of perceptual salience. More-over, the additional finding of valence-specific effects of differentcolors indicates that emotional and perceptual salience can interactto produce memory enhancement. One interesting question, how-ever, is whether differential effects of colors would still be presentin situations where color is a perceptually less salient signal.Findings on the effects of red in achievement contexts indicate thatred can evoke fear of failure, even when presented as a rathersubtle cue (Elliot, Maier, Moller, Friedman, & Meinhardt, 2007).These findings suggest that red and green could also differentiallyenhance emotional memory, even when color is not particularlysalient. However, to the extent that valence-specific effects ofdifferent colors reflect their conventional use in everyday life tosignal potential threats or benefits, it may also be that thesedifferential effects are only found when colors are experienced asimportant signals. Thus, the valence-specific effects of red andgreen on emotion-induced memory enhancement might vanishwhen employing memory paradigms that do not contain asignaling effect, such as presenting lists of words in whicheach word is shown in a different color. Indeed, determining therole of salience in the interplay between color and emotionalsignificance would be an important direction for future re-search.

Our findings suggest that red primarily carries negative meaningand that green primarily carries positive meaning. There may besome boundary conditions, however. First, it remains to be shownwhether the effects of red and green generalize across cultures.Although red and green are similarly used in a variety of cultures,it may be that effects of color depend on the color coding conven-tions of a culture. However, at least for red, there is evidence thatthere might be even a biological-based predisposition to interpretred as a danger signal (e.g., Khan, Levine, Dobson, & Kralik,2011), which would suggest that similar effects of colors might beobserved across different cultures. Second, recent findings suggestthat, in specific contexts, red can also carry positive meanings, forinstance, in the context of sexual attraction (e.g., Elliot & Niesta,2008). Thus, it might be that there are specific types of positivewords that also benefit from red, such as, for example, erotic

words. Indeed, exploring the context specificity of color effects isan important avenue for future research.

Our findings may also have more general implications for theuse of colors in experimental research on emotioncognition in-teractions. In several experimental paradigms, colors are tradition-ally used in arbitrary ways to signify specific conditions. Forinstance, in experiments on voluntary forgetting of emotionalmemories, red has typically been used as a forget cue, whereasgreen has been used as a remember cue (e.g., Depue, Banich, &Curran, 2006; Murray, Muscatell, & Kensinger, 2011). Given thatdifferent colors can have differential effects on emotional process-ing, color type might represent a confounding variable in studiesthat did not control for color effects. Finally, the results of thepresent study may be of considerable importance for applied fieldsas well. Current guidelines for choosing colors to effectivelyvisualize information refer to general stimulus features such asdistinctiveness or contrast (Ware, 2004). Our findings suggest thatit is important to also take differential effects of specific colors fordifferent types of information into account. Further exploring sucheffects will be an important avenue for future research.

ReferencesBower, G. H. (1981). Mood and memory. American Psychologist, 36,

129148. doi:10.1037/0003-066X.36.2.129Depue, B. E., Banich, M. T., & Curran, T. (2006). Suppression of emo-

tional and nonemotional content in memory. Psychological Science, 17,441447. doi:10.1111/j.1467-9280.2006.01725.x

Duyck, W., Desmet, T., Verbeke, L. P. C., & Brysbaert, M. (2004).WordGen: A tool for word selection and nonword generation in Dutch,English, German, and French. Behavior Research Methods, Instruments& Computers, 36, 488499. doi:10.3758/BF03195595

Elliot, A. J., & Aarts, H. (2011). Perception of the color red enhances theforce and velocity of motor output. Emotion, 11, 445449. doi:10.1037/a0022599

Elliot, A. J., Maier, M. A., Moller, A. C., Friedman, R., & Meinhardt, J.(2007). Color and psychological functioning: The effect of red onperformance attainment. Journal of Experimental Psychology: General,136, 154168. doi:10.1037/0096-3445.136.1.154

Elliot, A. J., & Niesta, D. (2008). Romantic red: Red enhances mensattraction to women. Journal of Personality and Social Psychology, 95,11501164. doi:10.1037/0022-3514.95.5.1150

Elliot, A. J., Payen, V., Brisswalter, J., Cury, F., & Thayer, J. F. (2011). Asubtle threat cue, heart rate variability, and cognitive performance.Psychophysiology, 48, 1340 1345. doi:10.1111/j.1469-8986.2011.01216.x

Hamann, S. (2001). Cognitive and neural mechanisms of emotional mem-ory. Trends in Cognitive Sciences, 5, 394400. doi:10.1016/S1364-6613(00)01707-1

Hunt, R. R. (1995). The subtlety of distinctiveness: What von Restorffreally did. Psychonomic Bulletin & Review, 2, 105112. doi:10.3758/BF03214414

Jacobs, K. W., & Hustmyer, F. E. (1974). Effects of four psychologicalprimary colors on gsr, heart rate, and respiration rate. Perceptual andMotor Skills, 38, 763766.

Khan, S. A., Levine, W. J., Dobson, S. D., & Kralik, J. D. (2011). Redsignals dominance in male rhesus macaques. Psychological Science, 22,10011003. doi:10.1177/0956797611415543

Meier, B. P., Robinson, M. D., & Clore, G. L. (2004). Why good guys wearwhite. Psychological Science, 15, 8287. doi:10.1111/j.0963-7214.2004.01502002.x

Michel, J. B., Shen, Y. K., Aiden, A. P., Veres, A., Gray, M. K., TheGoogle Books Team, . . . Aiden, E. L. (2011). Quantitative analysis of

This

docu

men

tis

copy

right

edby

the

Am

eric

anPs

ycho

logi

calA

ssoc

iatio

nor

one

ofi

tsal

lied

publ

isher

s.Th

isar

ticle

isin

tend

edso

lely

fort

hepe

rson

aluse

oft

hein

divi

dual

use

ran

dis

not

tobe

diss

emin

ated

broa

dly.

378 KUHBANDNER AND PEKRUN

culture using millions of digitized books. Science, 331, 176182. doi:10.1126/science.1199644

Moller, A. C., Elliot, A. J., & Maier, M. A. (2009). Basic hue-meaningassociations. Emotion, 9, 898902. doi:10.1037/a0017811

Morris, S. B., & DeShon, R. P. (2002). Combining effect size estimates inmeta-analysis with repeated measures and independent-group designs.Psychological Methods, 7, 105125. doi:10.1037/1082-989X.7.1.105

Murray, B. D., Muscatell, K. A., & Kensinger, E. A. (2011). Effects ofemotion and age on performance during a think/no-think memory task.Psychology and Aging, 26, 940955. doi:10.1037/a0023214

Okubo, M., & Ishikawa, K. (2011). Automatic semantic association betweenemotional valence and brightness in the right hemisphere. Cognition andEmotion, 25, 12731280. doi:10.1080/02699931.2010.541658

Reisberg, D., & Heuer, F. (2004). Memory for emotional events. In D.Reisberg and P. Hertel (Eds.), Memory and emotion (pp. 341). NewYork, NY: Oxford University Press. doi:10.1093/acprof:oso/9780195158564.003.0001

Travis, D. (1991). Effective color displays: Theory and practice. London,UK: Academic Press.

von Restorff, H. (1933). U ber die Wirkung von Bereichsbildungen imSpurenfeld [The effect of organization in perceptual fields]. Psycholo-gische Forschung, 18, 299342. doi:10.1007/BF02409636

Ware, C. (2004). Information visualization: Perception for design (2nded.). San Francisco, CA: Morgan Kaufman.

Wilson, G. D. (1966). Arousal properties of red versus green. Perceptualand Motor Skills, 23, 942949.

Appendix

English Translation of the Original German Emotional Words

Received June 18, 2012Revision received December 17, 2012

Accepted December 27, 2012

Neutral Negative Positive

Circle Crisis FriendEagle Fear FunFabric Jail HolidayFoil Misery KissesKiosk Nazi LaughPen Tumor LuckSign War PartyTable Weapon Winner

Note. In German language, the word length of all words ranged from five to six letters.

This

docu

men

tis

copy

right

edby

the

Am

eric

anPs

ycho

logi

calA

ssoc

iatio

nor

one

ofi

tsal

lied

publ

isher

s.Th

isar

ticle

isin

tend

edso

lely

fort

hepe

rson

aluse

oft

hein

divi

dual

use

ran

dis

not

tobe

diss

emin

ated

broa

dly.

379COLOR AND EMOTIONAL MEMORY

Joint Effects of Emotion and Color on MemoryMethodParticipantsMaterials and Procedure

ResultsEmotional RatingsEffects of Color and Emotional Valence

DiscussionReferences

Appendix English Translation of the Original German Emotional Words