New Displays and New Emotions: A Commentary on Rozin andCohen (2003)

Dacher Keltner and Michelle N. ShiotaUniversity of California, Berkeley

In this article, the authors elaborate on 3 ideas advanced in P. Rozin and A. B.Cohens (2003) innovative study of facial expression. Taking a cue from theirdiscovery of new expressive behaviors (e.g., the narrowed eyebrows), the authorsreview recent studies showing that emotions are conveyed in more channels thanusually studied, including posture, gaze patterns, voice, and touch. Building on theirclaim that confusion has a distinct display, the authors review evidence showingdistinct displays for 3 self-conscious emotions (embarrassment, shame, and pride),5 positive emotions (amusement, desire, happiness, love, interest), and sympathyand compassion. Finally, the authors offer a functional definition of emotion tointegrate these findings on new displays and emotions.

The contemporary study of emotion began, in onesense, with a rather simple, and for many, uncontro-versial finding: People in different parts of the worldinterpret a limited number of facial expressions ofemotion in similar ways. This simple assertion, whichemerged in the well-known studies of Ekman, Frie-sen, and Izard in the late 1960s and early 1970s, hassince spawned a variety of questions and debates (forreview, see Keltner, Ekman, Gonzaga, & Beer, inpress). To what extent is the recognition of facialexpression actually universal? Does facial expressionindeed signal emotion, social intention, or some com-bination of these kinds of inferences? And whichemotions possess a recognizable signal?

Rozin and Cohens article (2003), at first blush,raises an interesting possibility related to this lastquestion: Does confusion have a recognizable signal?They turned 255 introductory psychology studentsinto participant observers and asked them to identify,via target reports, the emotions that others expressedin the flow of spontaneous interaction. The findingthat struck Rozin and Cohen, and which strikes us, isthe frequency with which participants reported facialcommunication of emotions not recognized by most

emotion taxonomies in our field. These include suchemotions as confusion, concentration, worry, flirta-tiousness, skepticism, and indifference. This findingparallels work done in other domains of emotion. Forexample, if you ask participants to list states theyconsider emotions, odds are love would emergeamong the most frequent responses (Fehr & Russell,1984). By contrast, love rarely makes emotion theo-rists taxonomies (but see Shaver, Schwartz, Kirson,& OConnor, 1987).

In this brief commentary, we rely on Rozin andCohens (2003) innovative methods and suggestivefindings as a platform to consider three broader ques-tions. First, what does the face reveal? Second, howmany emotions indeed have distinct signals? And fi-nally, what is an emotion?

How Are Emotions Communicated?

One of the most widespread critiques of the studyof emotional expression is that it has focused on ex-treme, if not exaggerated, displays of emotion (e.g.,Russell, 1994). This methodological emphasis wasnot accidental but based on a clear theoretical ratio-nale: the coevolution of human emotional display andthe capacity to perceive those displays produced pro-totypical displays of emotion. Nonetheless, this em-phasis on prototypical displays has generated an em-pirical literature that is relatively silent with respect tothe more subtle, varied, complex, and perhaps evenidiosyncratic ways in which people express most oftheir emotions in real life.

Dacher Keltner and Michelle N. Shiota, Department ofPsychology, University of California, Berkeley.

Correspondence concerning this article should be ad-dressed to Dacher Keltner, Department of Psychology,University of California, 3319 Tolman Hall, Berkeley, Cali-fornia 94720-1650. E-mail: keltner@socrates.berkeley.edu

Emotion Copyright 2003 by the American Psychological Association, Inc.2003, Vol. 3, No. 1, 8691 1528-3542/03/$12.00 DOI: 10.1037/1528-3542.3.1.86

86

On this point Rozin and Cohen (2003) providedsome very germane evidence. In analyzing partici-pants nave accounts of the facial behavior associatedwith different emotions, they document some inter-esting, as yet undocumented, displays of differentstates (see Rozin & Cohen, 2003, Table 3). Narrowedeyes accompany the experience and social display ofconfusion; lip bites accompany worry; and the fur-rowed or lowered brow accompanies concentration,as if to signal that the person is immersed in thoughtand not to be disturbed. The take-home message: Re-gions of the face and relatively discrete actions cansignal more complex states much like the more com-plex, prototypical facial muscle configurations thathave been well-studied.

Clearly more evidence is needed before the fieldcan think of narrowed eyes as signs of confusion or ofthe lip bite as a sign of anxiety. It is important to showthat these expressive behaviors actually correlate withother measures of affective experience, such as self-report or autonomic response. More precise definitionand coding of the facial behaviors themselves wouldbe required (note that narrowed eyes could resultfrom the movement of several facial muscles, includ-ing Facial Action Coding System Action Units 4, 6,and 7). It would also be important to show that thesefacial behaviors signal the specific states named andnot others. It would be important to show whether thefacial signals described demonstrate the dynamicproperties of affective displaysin particular, a du-ration in the vicinity of 15 s. Finally, to the degreethat the status of these states as universal basic emo-tions is of interest, cross-cultural work regarding theuniversality of display and recognition is needed. Thiswould lend more confidence to the assertion that theseare affective displays and not other kinds of displayssuch as emblematic displays or referential expressions(e.g., Ekman, 1993).

More generally, these findings make the importantpoint that complex states can be signaled by relativelyisolated, simple actions such as narrowed eyes or lipbites. For example, in many parts of Southeast Asia,people can convey shame with a tongue bite (Haidt &Keltner, 1999). In other work, Rozin and his col-leagues have found that different variants of the dis-gust and contempt expressions actually communicateslightly different meanings (Rozin, Lowery, & Ebert,1994).

Taking this implication one step further, Rozin andCohens (2003) study encourages the field to thinkmore broadly about the ways in which emotion iscommunicated. One obvious candidate is posture.

Changes in head orientation and expansion or con-traction of the chest were of interest to Darwin (1872),among others, and are certain to communicate differ-ent emotions. For example, postural constriction ispart of the signal of embarrassment and shame (Kelt-ner, 1995; Keltner & Harker, 1998). Our own recentwork suggests that postural expansion is involved inthe pride display (Shiota, Campos, & Keltner, inpress), and forward leans communicate romantic love(Gonzaga, Keltner, Londahl, & Smith, 2001).

Another candidate is touch. Darwin (1872) recog-nized early on that touch was the most universaldemonstration of intimacy across species, and theimportance of touch in social bonding is well-established (e.g., Klaus & Kennell, 1983; Shaver,Morgan, & Wu, 1996). This effect may be based, inpart, on physiological effects of social touch suchas decreased sympathetic arousal and increased vagaltone (Uvnaes-Moberg, 1997). Touch can take onmany forms, and when one considers variability in thelocation of the touch, type of pressure, and duration,there is room for information about discrete emotionalexperience as well as intensity (Hertenstein, 2002).

Although neglected, the vocal communication ofhuman emotion has received some empirical study(Scherer, 1986). More work in this area has examinednonhuman primate communication, finding, for in-stance, that distinct vocal calls are associated withaffiliative behavior, mating, and the location of food(Snowden, in press). Even odor may play a role in thecommunication of emotion. In his review of signaling,Snowden discussed ways in which common affectlikestates are communicated through scent in nonhumanspecies (Snowden, in press). The same may be true ofhumans.

These are all very speculative possibilities. Yet, wethink they are fruitful ones and likely to generatenovel insights about emotion if researchers pursuemore open-ended methodologies, focus on differentparts of the face and body, and look to new states, asRozin and Cohen (2003) have done. This kind of re-search is likely to reveal that humans express emotionin a far more varied and continuous fashion than iscurrently assumed by the field.

How Many Emotions Have Signals?

The field of emotion has been profoundly shapedby the study of facial expression. Many taxonomies ofemotion hinge on the states that have universally re-liable signals. In a more practical sense, the ability tomeasure several facial expressions in spontaneous be-

COMMENTARIES 87

havior and to represent emotion in photographs hasplayed a significant role in the study of emotion-related autonomic response and central nervous sys-tem activity. When new facial expressions of emotionare documented, many things are at stake, from theo-retical taxonomies to potential advances in other areasto questions regarding classes of emotion (e.g., arethere emotions with no signal as well as emotionswith a defined signal?).

The Rozin and Cohen (2003) evidence suggeststhat confusion may have its own distinct signal. Wehave already noted the need for further evidence tobolster this claim. More generally, however, Rozinand Cohens findings raise an interesting question:How many states have distinct signals? Recent studiessuggest that more emotions may be expressed thanpreviously thought. A selective review of the litera-ture finds distinct displays for embarrassment andshame (Keltner, 1995; Keltner & Harker, 1998). Em-barrassment is marked by a sequence of gaze aver-sion, controlled smile, head turn (which exposes theneck), and occasional face touch (in other studies, justthe face touch is sufficient to communicate embar-rassment; Haidt & Keltner, 1999). Shame is displayedin a coordinated sequence of downward gaze and headmovements lasting 15 s. Discoveries of these dis-plays have led researchers to relate reduced embar-rassment to increased externalizing troubles in ado-lescent males (Keltner, Moffitt, & Stouthamer-Loeber, 1995) and increased shame to theunwillingness to disclose a past history of sexualabuse (Bonanno et al., 2002). Studies are also findinga distinct display of pride, as evident in postural ex-pansion and upwards head and gaze movements(Shiota et al., in press). At least three self-consciousemotions appear to have distinct displays.

In the realm of positive emotions, it is often as-sumed that there is one positive emotionhappi-nessand by implication, one display of positiveemotion, the Duchenne smile, which involves the ac-tion of the orbicularis oculi (Ekman, 1993). Althoughit may be true that the Duchenne smile is common toexpressions of many positive emotions, recent studiesfind that other facial actions differentiate the displaysof positive emotion. For example, the momentary ex-perience of love is expressed in a coherent pattern ofsmiling, mutual gaze, open postures, and forwardleans (Gonzaga et al., 2001). Desire is signaled in avariety of lip-related actions, including lip licks,wipes, and tongue protrusions (Gonzaga, Turner,Keltner, Altemus, & Campos, 2002). And there issome evidence to suggest that laughter and smiling

have distinct experiential correlates and social conse-quences (Keltner & Bonanno, 1997). These findingssuggest that at least five positively valenced stateslove, desire, amusement, happiness, and interest(Reeve, 1993)have distinct displays. In our ownresearch, we have shown that awe and amusement, inthis case posed rather than experienced, have distinctactions (Shiota et al., in press).

Finally, Eisenberg and colleagues have collectedimpressive evidence indicating that a display of sym-pathy, namely oblique eyebrows and concerned gaze,is correlated with increased sympathy, heart rate de-celeration, and increased helping behavior, and thatthis display is different from that of distress (Eisen-berg et al., 1989). Ensuing judgment studies havefound that posed displays of sympathy are not judgedwith a great deal of reliability, however (Haidt &Keltner, 1999). It should be noted that the emotions inthe original list of basics have what Ekman has calledsnapshot qualities, in that a single moment in timeat the apex of the expression is sufficient for decod-ing, whereas these other displays involve a flow ofmovement over time (Ekman, 1993).

What Is an Emotion Anyway?

Our cursory review suggests that humans conveymany more emotions through facial behavior andother signals than previously assumed. As researchersopen the door, as Rozin and Cohen (2003) have done,to new, or at least newly recognized emotions, theyraise perhaps the most vexing question of all: Whatexactly is an emotion? Are we justified in callingconfusion, concentration, and worry emotions? Inasking this kind of question, the field refines its con-ceptualization of emotion.

As Rozin and Cohen (2003) suggested, definitionsand taxonomies of emotion tend to derive from re-searchers particular methods. For those guided by thestudy of facial expression, the basic or primary emo-tions are the six tested in the early facial expressionstudies of Ekman, Friesen, and Izard: happiness, sur-prise, anger, sadness, fear, and disgust. Lists based onappraisal research tend to include more complex, self-conscious emotions such as shame and guilt (e.g.,Scherer, 1997). Lists based on animal models ofphysiology tend to include more positive emotions,such as lust and maternal love, than most (e.g., Pank-sepp, 1998). Lists based on emotion vocabulary orcore relational themes, as described in narratives,tend to include more social emotions, such as love andpride (e.g., Lazarus, 1991; Shaver et al., 1987).

COMMENTARIES88

For this reason, it is often productive to turn toemotion taxonomies of lay individuals or other cul-tures, as Rozin and Cohen (2003) have done, tobroaden our conceptualization of emotion. The list ofemotion words available to the participant observersin Rozin and Cohens study was based on pretest datafrom a similar set of observers, and therefore reflectsstates that lay observers consider facially communi-cable emotions.

In looking at this extensive list of emotions, it isnoteworthy how many emotions involve reactions tothe individuals understanding or state of knowledgevis-a`-vis the environment. There seems to be a familyof emotions, epistemology-based emotions, that arisein response to the nature of the individuals world-view. Confusion and concentration, in particular, areprimarily reactions to informationconfusion is thefeeling that the environment is giving insufficient orcontradictory information, and concentration is thefeeling that the information provided by the environ-ment has reached the maximum level one can process.Worry also involves uncertainty about the possiblenegative status of a situation and calls for confirma-tion one way or the other.

Historically, little attention has been given to theseepistemological emotions (although see Oatley &Johnson-Laird, 1987). Other emotions that arise inresponse to the individuals understanding of the ex-ternal world include amusement, awe (Keltner &Haidt, in press), and interest (Reeve, 1993). Rozin andCohens (2003) findings suggest that these are com-mon states that have readily identifiable signals. Ofcourse, more evidence is needed regarding thesestates appraisal patterns and physiological profilesbefore the field can more confidently include themunder the rubric emotion.

More generally, what are emotion researchers tomake of the list that Rozin and Cohens (2003) par-ticipants produced? The following is our reproductionof their list to address that question:

afraid (fearful, anxious, apprehensive, scared); agree-ment; amused (giddy, humor, silly, joke, laugh, mischie-vous, secretive, sly, teasing); anger (annoyed, insult, ir-ritated); ashamed (embarrassed, sheepish, shy, timid);attracted; awed; awkward; bored (indifferent); concen-tration; concerned (worried); confident (proud); con-fused (consternation, puzzled, uncertain); content(fulfilled, pleased, relaxed, satisfied); pensive (contem-plative, thoughtful); contempt (disdain); cool; curious(inquisitive, interested, intrigued); desperate; disap-pointed; sad (unhappy); disapproving (dislike); discom-fort (pain); disgust (grossed out); envious (jealous); ex-cited; flirtatious (sexy, coy); frustrated; guilty; happy

(joyful); sarcastic (wry); surprised (shocked); skeptical;sympathetic; tired; upset; and worried (upset, nervous).

Clearly, there is a great deal of overlap with currentemotion taxonomies. There are also new states worthyof study, as we have noted. And certainly, researcherswould tend to recategorize these states into fewer cat-egories; for example, desperate, disappointed, and sadwould probably go into one category.

We note several trends in this list. Nearly all of thewords reflect transient states rather than traits. Mostwords reflect a reaction to some situational elicitoroutside the self. These elicitors fall into two broadcategories: social events and information. Finally,most of the words are evaluative and imply that someeffect on or action by the elicitor is desired. For in-stance, confusion is a negatively valenced state thatcan be alleviated if the environment provides moreinformation.

Whereas emotion taxonomies and definitions tendto emphasize intrapersonal functions of emotion (seeLevenson, 1999, for a discussion of this type of func-tion), this list suggests that a defining feature of emo-tion is its social extension, that is, how an inner stateinfluences the social environment. The valence of theemotion words used in this study implies a desire toinfluence or change the environment. With respect tothe states described here, displays of confusion maysay I dont understandtell me more. Displays ofworry may say I think somethings wrong out there,but Im not sureplease deny or confirm. Concen-tration displays may say Ive got all the informationI can handle right nowdont send any more yet.Emotions are functional, providing an impetus forchange and a mechanism by which to influence onessurroundings (Barrett & Campos, 1987; Keltner &Gross, 1999; Keltner & Haidt, 2001; Shiota, Campos,Keltner, & Hertenstein, in press).

In light of these considerations, we offer thefollowing definition: An emotion is a universal, func-tional reaction to an external stimulus event, tempo-rarily integrating physiological, cognitive, phenom-enological, and behavioral channels to facilitate afitness-enhancing, environment-shaping response tothe current situation. This definition reflects the con-tent of the operational list used by Rozin and Cohen(2003) as well as research from a range of method-ological paradigms.

The states on which Rozin and Cohen (2003) fo-cusedconfusion, worry, and concentrationfulfillpart, but not all, of this definition. All could be con-strued as reactions to information presented by the

COMMENTARIES 89

environmentexternal stimulus events. The functionof these epistemology-based emotions remains a bit ofa mystery. Studies of the antecedents, signals, andconsequences of confusion, worry, and concentrationwould be germane and are likely to reveal that thesestates have clear purposes after all.

References

Barrett, R., & Campos, J. (1987). Perspectives on emotionaldevelopment II: A functionalist approach to emotions. InJ. D. Osofsky (Ed.), Handbook of infant development(2nd ed., pp. 555578). New York: Wiley.

Bonanno, G. A., Keltner, D., Noll, J. G., Putnam, F. W.,Trickett, P. K., LeJeune, J., et al. (2002). When the facereveals what words do not: Facial expressions of emotion,smiling, and the willingness to disclose childhood sexualabuse. Journal of Personality and Social Psychology, 83,94110.

Darwin, C. (1872). The expression of the emotions in manand animals. London: John Murray.

Eisenberg, N., Fabes, R. A., Miller, P. A., Fultz, J., Shell,R., Mathy, R. M., et al. (1989). Relation of sympathy anddistress to prosocial behavior: A multimethod study.Journal of Personality and Social Psychology, 57, 5566.

Ekman, P. (1993). Facial expression and emotion. AmericanPsychologist, 48, 384392.

Fehr, B., & Russell, J. A. (1984). Concept of emotionviewed from a prototype perspective. Journal of Experi-mental Social Psychology: General, 113, 464486.

Gonzaga, G. C., Keltner, D., Londahl, E. A., & Smith,M. D. (2001). Love and the commitment problem in ro-mantic relations and friendship. Journal of Personalityand Social Psychology, 81, 247262.

Gonzaga, G. C., Turner, R., Keltner, D., Altemus, M., &Campos, B. (2002). Love and desire in intimate bonds.Manuscript submitted for publication.

Haidt, J., & Keltner, D. (1999). Culture and facial expres-sion: Open-ended methods find more expressions and agradient of recognition. Cognition and Emotion, 13, 225266.

Hertenstein, M. J. (2002). Touch: Its communicative func-tions in infancy. Human Development, 45, 7094.

Keltner, D. (1995). The signs of appeasement: Evidence forthe distinct displays of embarrassment, amusement, andshame. Journal of Personality and Social Psychology, 68,441454.

Keltner, D., & Bonanno, G. A. (1997). A study of laughterand dissociation: The distinct correlates of laughter andsmiling during bereavement. Journal of Personality andSocial Psychology, 73, 687702.

Keltner, D., Ekman, P., Gonzaga, G. C., & Beer, J. (in

press). Facial expression of emotion. In R. Davidson, K.Scherer, & H. Goldsmith (Eds.), Handbook of affectivescience. London: Oxford University Press.

Keltner, D., & Gross, J. J. (1999). Functional accounts ofemotion. Cognition and Emotion, 13, 467480.

Keltner, D., & Haidt, J. (2001). Social functions of emo-tions. In T. Mayne & G. Bonanno (Eds.), Emotions: Cur-rent issues and future directions (pp. 192213). NewYork: Guilford Press.

Keltner, D., & Haidt, J. (in press). Approaching awe, amoral, aesthetic, and spiritual emotions. Cognition andEmotion.

Keltner, D., & Harker, L. A. (1998). The forms and func-tions of the nonverbal display of shame. In P. Gilbert &B. Andrews (Eds.), Interpersonal approaches to shame(pp. 7898). Oxford, England: Oxford University Press.

Keltner, D., Moffitt, T., & Stouthamer-Loeber, M. (1995).Facial expressions of emotion and psychopathology inadolescent boys. Journal of Abnormal Psychology, 104,644652.

Klaus, M. H., & Kennell, J. H. (1983). Bonding: The begin-nings of parentinfant attachment. New York: Mosby,Inc.

Lazarus, R. S. (1991). Emotion and adaptation. New York:Oxford University Press.

Levenson, R. W. (1999). The intrapersonal functions ofemotion. Cognition and Emotion, 13, 481504.

Oatley, K., & Johnson-Laird, P. N. (1987). Towards a cog-nitive theory of emotions. Cognition and Emotion, 1, 2950.

Panksepp, J. (1998). Affective neuroscience: The founda-tions of human and animal emotions. New York: OxfordUniversity Press.

Reeve, J. (1993). The face of interest. Motivation and Emo-tion, 17, 353375.

Rozin, P., & Cohen, A. B. (2003). High frequency of facialexpressions corresponding to confusion, concentration,and worry in an analysis of naturally occuring facial ex-pressions of Americans. Emotion, 3, 6875.

Rozin, P., Lowery, L., & Ebert, R. (1994). Varieties ofdisgust faces and the structure of disgust. Journal of Per-sonality and Social Psychology, 66, 870881.

Russell, J. A. (1994). Is there universal recognition of emo-tion from facial expression? A review of the cross-cultural studies. Psychological Bulletin, 115, 102141.

Scherer, K. S. (1986). Vocal affect expression: A reviewand a model for future research. Psychological Bulletin,99, 143165.

Shaver, P. R., Morgan, H. J., & Wu, S. (1996). Is love abasic emotion? Personal Relationships, 3, 8196.

Shaver, P., Schwartz, J., Kirson, D., & OConnor, C.(1987). Emotion knowledge: Further exploration of a pro-

COMMENTARIES90

totype approach. Journal of Personality and Social Psy-chology, 52, 10611086.

Shiota, M. N., Campos, B., & Keltner, D. (in press). Thefaces of positive emotion: Prototype displays of awe,amusement, and pride. Annals of the New York Academyof Sciences.

Shiota, M. N., Campos, B., Keltner, D., & Hertenstein, M. J.(in press). Positive emotion and the regulation of inter-personal relationships. In P. Philippot & R. S. Feldman(Eds.), The regulation of emotion. Mahwah, NJ: Erlbaum.

Snowden, C. (in press). Animal communication of emotion.

In R. Davidson, K. Scherer, & H. Goldsmith (Eds.),Handbook of affective science. London: Oxford Univer-sity Press.

Uvnaes-Moberg, K. (1997). Physiological and endocrine ef-fects of social contact. In C. S. Carter, I. I. Lederhendler,et al. (Eds.), The integrative neurobiology of affiliation(pp. 146163). New York: Academy of Sciences.

Received October 16, 2002Revision received November 14, 2002

Accepted November 14, 2002

COMMENTARIES 91