Positive and Negative: Infant Facial Expressions and … dominant theory holds that ... discrete emotions, such as joy and anger (Ekman, 1994). Each ... Positive and Negative: Infant

Copyright © 2002 American Psychological Society

1

CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE

cause infants cannot use words totell us directly about their emo-tions. In this article, I report on newattempts to understand infants’ fa-cial expressions by investigatingboth the situations in which differenttypes of expressions take place andnaive observers’ ratings of the emo-tional intensity of those expres-sions. I conclude by describing in-novative techniques that promise toilluminate long-standing issues inthe communication of emotion.

A DOMINANT THEORYAND ITS VICISSITUDES

One dominant theory holds thatthere are a relatively small numberof basic, discrete emotions, such asjoy and anger (Ekman, 1994). Eachinvolves a distinct emotion pro-gram that produces a prototypicfeeling state and facial expression.Infancy provides crucial informa-tion about the relationship betweenemotion and facial expression be-cause infants are not affected by so-cial rules concerning when it is ap-propriate and not appropriate tosmile or show other expressions.Do infants typically display dis-crete expressions of positive andnegative emotion as predicted?

It is difficult to elicit discretenegative expressions, such as ex-pressions of anger, distress, andsadness, in infants. Blends of theseexpressions are common (Matias &Cohn, 1993), and infants tend to cy-cle through anger and distress ex-pressions during a single cryingbout (Camras, 1992). In fact, thecry-face is the most common infant

Abstract

One path to understandingemotional processes and theirdevelopment is the investiga-tion of early facial expressions.Converging evidence suggeststhat although all infant smilesindex positive emotion, somesmiles are more positive thanothers. The evidence stemsboth from the situations inwhich infants produce differ-ent facial expressions and fromnaive observers’ ratings of theemotional intensity of the ex-pressions. The observers’ rat-ings also suggest that similarfacial actions—such as cheekraising—lead smiles to be per-ceived as more positive andlead negative expressions (cry-faces) to be perceived as morenegative. One explanation forthis parsimony is that certainfacial actions are associatedwith the intensification of bothpositive and negative emo-tions.

Keywords

emotion; facial expression; in-fancy; joy; distress

Extreme expressions—from thefierce grimace to the crestingsmile—characterize early humanlife. These expressions often elicitstrong emotions in persons close toan infant and play a role in forminginteractive patterns that influencelater development. But our under-standing of the link between in-fants’ facial expressions and theiremotional experience is limited be-

reaction to all negative situations.In the cry-face, the brows are low-ered and the lip corners are pulledto the side; there is a variable de-gree of mouth opening and cheekraising (Oster, 2000; see Fig. 1). Thecry-face incorporates features ofboth anger and distress expres-sions and may index a shared coreof negative emotionality. Naivejudges, for example, rate discreteanger and distress expressions(variants of the cry-face) as show-ing high levels of both anger anddistress (Oster, Hegley, & Nagel,1992).

Smiles, in contrast, are easilyrecognized by naive judges as dis-crete expressions of joy in infancyand beyond. However, adultssometimes smile in unpleasant sit-uations or simply in response to so-cial interchange. The dominanttheory has addressed this inconsis-tency by postulating a distinctionbetween joyful and nonjoyful typesof smiling (Ekman, 1994). Cheek-raise, or Duchenne, smiles, in whichthe muscle around the eye con-tracts, lifting the cheeks high, arethought to be felt expressions of joy(left-hand column of Fig. 2). Smileswithout cheek raising are thoughtto be nonemotional signals used tolubricate social interactions andmask negative feeling (right-handcolumn of Fig. 2).

The distinction between differ-ent types of smiling is supportedby a study in which adults showedmore cheek-raise smiling whileviewing amusing films than whileviewing gruesome films, and thequantity of cheek-raise smiling wasassociated with self-reported posi-tive emotion (Ekman, Davidson, &Friesen, 1990). Smiles withoutcheek raising, by contrast, did nottend to occur during the amusingfilm, nor were they associated withself-reported positive emotion.Even among 10-month-old infants,cheek-raise Duchenne smilestended to occur in reaction tomother’s smiling approach, and

Positive and Negative: Infant Facial Expressions and Emotions

Daniel S. Messinger

1

Department of Psychology, University of Miami, Coral Gables, Florida

2 VOLUME 11, NUMBER 1, FEBRUARY 2002

Published by Blackwell Publishing Inc.

gel, & Dickson, 1999). A rise or fallin an infant’s cheek-raise smilingduring a given interaction with hisor her mother tended to be mir-rored by a rise or fall in the infant’ssmiling without cheek raising.Most tellingly, individual smileswith cheek raising tended to bepreceded by smiles without cheekraising. This suggests that infantsfrequently begin a smile withoutcheek raising that involves a con-strained degree of positive emo-tion, and that this emotion then in-tensifies, and is reflected in a morepositive smile with cheek raising.

The dominant theoretical per-spective has not countenanced thesuggestion that smiles with andwithout cheek raising are distin-guished only by quantitative differ-ences in the intensity of positiveemotion. In focusing on cheek-raisesmiling as a unique index of posi-tive emotion, the theory has alsoneglected the emotional signifi-cance of open-mouth (play) smilingin which the jaw is dropped (upperhalf of Fig. 2). Open-mouth smilingis common and tends to occur withcheek-raise smiling (upper left-hand quadrant of Fig. 2), especiallyduring social games. Is there, then,a single discrete expression of joy?

DIFFERENT SMILES ARE PRODUCED IN DIFFERENT

SITUATIONS

The situations in which younginfants smile suggest that all smilesare emotionally positive, but spe-cific types of smiling are more posi-tive than others (Messinger, Fogel,& Dickson, 2001). If only cheek-raisesmiling is joyful, smiling alone—which does not involve cheek rais-ing or mouth opening—should notoccur in periods of interaction ex-pected to elicit positive emotion. Arecent study, however, found thatsmiling alone, expressed as a pro-portion of time without smiling,

smiles without cheek raises tendedto occur in reaction to the approachof an impassive stranger (Fox &Davidson, 1988).

Research with 1- to 6-month-oldinfants showed, however, thatsmiles with and without cheekraising are related (Messinger, Fo-

Fig. 1. Mean upset-distress ratings (on a scale from 0 to 8) and their standard errorsfor each pair of cry-faces. The cry-faces on the left involve cheek raising, and the cry-faces on the right involve only very week cheek raising. The expressions in each hor-izontal pair were created with the identical cry-face mouth. In each column, themouth is open in the top two cry-faces and closed in the bottom two. Pairs of expres-sions also differ in the strength of the cry-faces portrayed.


CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 3

cheek-raise smiling when theirmothers were smiling rather thannot smiling. They engaged in moreopen-mouth smiling when gazingat than away from their mothers’faces. Infants engaged in morecombined open-mouth cheek-raisesmiling than smiling alone bothwhen their mothers were smilingand when they were gazing at theirmothers (Fig. 3).

Different types of smiling appearto involve the intensification of dif-ferent positive processes that arepresent to a lesser degree in smilingalone (Messinger et al. , 2001).Cheek-raise smiling was especiallyassociated with the reciprocation ofpositive affect, whereas open-mouth smiling was especially as-sociated with visual engagement.Between 1 and 6 months of age,infants’ open-mouth cheek-raisesmiling became more prevalentwhen the infants were gazing attheir mothers’ faces while the moth-ers were smiling. This suggests thatas infants develop, they become in-creasingly likely to use open-mouthcheek-raise smiling to participate inpeak positive interchanges in whichjoy is shared during mutual engage-ment (Fogel, Nelson-Goens, Hsu, &Shapiro, 2000).

These results suggest that there isno single smile of joy in infancy. In-stead, all smiles are linked to posi-tive social processes, and differenttypes of smiling appear to be spe-cialized for different types of posi-tive interchanges. One interpreta-tion is that all smiles are positive butthat some smiles are more positivethan others. Do the ratings of naiveobservers support this inference?

ADULT PERCEPTIONSOF INFANTS’ SMILES

AND CRY-FACES

My colleagues and I recently de-signed a study to investigatewhether all smiles are perceived as

tended to occur both when infantswere gazing at their mothers’ facesand when their mothers were smil-ing (Fig. 3). Smiling alone, then, ap-peared to be involved both in visualengagement with mother (gazing)and in sharing positive affect (mu-tual smiling; Messinger et al., 2001).

If even smiling alone tended tooccur in positive periods of inter-action, when did other types ofsmiling occur? Cheek-raise andopen-mouth smiling were each ex-pressed as proportions of timesmiling alone to answer this ques-tion. Infants engaged in more

Fig. 2. Mean joy-happiness ratings (on a scale from 0 to 8) and their standard errorsfor each pair of smiles. The smiles on the left are cheek-raise (Duchenne) smiles, andthe smiles on the right are not. The expressions in each horizontal pair were createdwith the identical smiling mouth. In each column, the top two smiles are open-mouthsmiles, and the bottom two are not. Pairs of expressions also differ in the strength ofthe smiles portrayed.



positive and whether smiles withcheek raising and smiles with mouthopening are perceived as more posi-tive than smiles without those fea-tures. The study included a parallelinvestigation of negative perceptionsof cry-faces, because there is circum-stantial evidence that cry-face ex-pressions involving cheek raisingand mouth opening are associatedwith more intense negative emo-tion than cry-faces without thosefeatures (Fox & Davidson, 1988).

For the rating study, a photo-editor was used to create identicalpairs of smiles and pairs of cry-faces that differed only in cheekraising (see Figs. 1 and 2).

2

Differ-ent pairs showed different degreesof mouth opening and strength ofthe underlying expression. Eachsmile and cry-face was shown, in arandomized order, with a compari-son neutral expression to 50 under-graduate participants who ratedthe happiness and joy (positive

emotion) they perceived in the dif-ferent types of smiles and the dis-tress and upset (negative emotion)they perceived in the cry-faces.

All smiles were perceived asmore emotionally positive than thecomparison neutral expression. Allcry-faces were perceived as moreemotionally negative than the com-parison neutral expression. Evensmiles and cry-faces involvingneither cheek raising nor mouthopening (see the lower right-hand

Fig. 3. Infant smiles as a function of whether the mother was smiling and whether the infant was gazing at the mother. In the bot-tom graph, smiling alone is expressed as a proportion of time smiling alone plus time without smiling. In each of the three graphsat the top, the indicated smiling type is expressed as a proportion of that smiling type plus smiling alone. Standard error bars sur-round each mean. From “All Smiles Are Positive, but Some Smiles Are More Positive Than Others,” by D.S. Messinger, A. Fogel,and K.L. Dickson, 2001, Developmental Psychology, 37, p. 647. Copyright 2001 by the American Psychological Association. Adaptedwith permission.


CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 5

quadrants in Figs. 1 and 2) were per-ceived, respectively, as more positiveand more negative than neutral ex-pressions. These results suggest thatall types of infant smiling communi-cate joy and that all cry-faces com-municate distress. What features ofthe expressions, then, were associ-ated with perceptions of more in-tense positive and negative emotion?

Smiles with cheek raising (Du-chenne smiles) were perceived asmore emotionally positive thansmiles without cheek raising. Cry-faces with cheek raising (and fore-head knotting) were perceived asmore emotionally negative thancry-faces without. (See the graphs inFigs. 1 and 2.) Mouth opening wasalso associated with ratings ofgreater emotional intensity for bothsmiles and cry-faces. Smiles withmouth opening were perceived asmore positive than smiles without.Cry-faces with mouth opening wereperceived as more negative thancry-faces without. The cry-faceswith mouth opening also tended tobe stronger than cry-faces withoutmouth opening. In fact, strongersmiles and cry-faces were also bothperceived as more emotionally in-tense than weaker versions of theseexpressions. The relative heights ofthe graphed lines in Figures 1 and 2illustrate these differences.

EXPRESSION FAMILIES—EMOTION FAMILIES

The ratings study paralleled ear-lier findings in indicating that allsmiles are perceived positively, butsome more positively than others.Similarly, all cry-faces were per-ceived negatively, but some morenegatively than others. It may bethat different types of smiles aredynamically related members of afamily of expressions, as are differ-ent types of cry-faces. The smileand cry-face expression families arelinked to corresponding emotion

families, each of which is distin-guished by a common motivationalorientation to the environment. Dif-ferent types of smiles play a role ina family of related positive emo-tions distinguished by a desire tocontinue pleasant experience. Dif-ferent cry-faces play a role in a fam-ily of related negative emotions dis-tinguished by a desire to stopunpleasant experience.

Specific features of these facialexpressions provide clues to distin-guishing the qualities of relatedemotions within emotion families.Continued research on the situa-tions in which cry-faces and smilesoccur, as well as additional ratingstudies using larger samples of in-fants with more diverse samples ofexpressions, are necessary. It nev-ertheless appears likely that similarfacial features are markers of bothemotionally intense positive andemotionally intense negative ex-pressions. What would account forthis parsimony?

Intensity, arousal, and attentiondimensions might distinguish dif-ferent types of smiles and cry-faces. It makes sense that strongerfacial expressions are especiallyemotionally intense. If a given feel-ing is associated with the contrac-tion of certain muscles, it is not sur-prising that more intense instancesof the feeling would be associatedwith stronger contraction of thosemuscles. Mouth opening may beassociated with increased arousalin both smiles and cry-faces. Forsmiles, the arousal is often sociallyelicited, and the smiles appear tohave an unconstrained, immediatequality. In cry-faces, mouth open-ing may be more directly linked tothe arousal associated with crying.Cheek raising, in contrast, reducesthe visual field and may be linkedto a feeling—and the object of thatfeeling—becoming more focal. Insmiles, cheek raising may be tied toa rising tide of pleasure. This maytypically occur when sharing apositive experience and so commu-

nicate one’s involvement in that ex-perience to a partner. In cry-faces,the strengthening of cheek raisingmay be linked to the transitionfrom anger to distress, in whichfeelings of trying to overcome anobstacle change to feelings of beingoverwhelmed by discomfort.

NEW DIRECTIONS

Several enduring questions re-main unresolved. Is there a one-to-one correspondence betweeninfants’ emotional feeling and ex-pression? Specific types of infantsmiling tend to occur in specific sit-uations, but do not typically pre-dominate in those situations. Cry-faces—not expressions of discretenegative emotions—are infants’most likely reaction to all negativesituations. These findings suggestthat infants’ feelings are usuallybut not always elicited in the ex-pected situations. It may also bethe case that infants’ feelings andfacial expressions are themselvesprobabilistically associated. It islikely that facial expressions influ-ence as well as express feeling andthat these emotional processes arethemselves intimately tied to par-ticipation in ongoing interactions.Two new tools may help research-ers deepen their understanding ofthe reciprocal influence of expres-sions, feelings, and interactive situ-ations.

New software should allow pre-cise quantitative measurement offacial action in time (Cohn, Zlo-chower, Lien, & Kanade, 1999).This may shed light on how ex-pressions change during an ongo-ing interaction, becoming weakerand stronger, and involving moreand less cheek raising and mouthopening before they fade. As astart, such software might confirmthat strong smiles and cry-facestend to co-occur with both cheekraising and mouth opening, creat-



adults arise? Detailed descriptionsof emotional expression will be animportant path to understandingcontinuity and development inemotional functioning through thelife span.

Recommended Reading

Camras, L.A. (1992). (See References)Cohn, J. (2000).

Automated face analy-sis

. Retrieved August 23, 2001,from Carnegie Mellon UniversityWeb site: http://www-2.cs.cmu.edu/~face/

Izard, C.E., & Ackerman, B.P. (2000).Motivational, organizational, andregulatory functions of discreteemotions. In M. Lewis & J.M.Haviland-Jones (Eds.),

Handbookof emotions

(2nd ed., pp. 253–264).New York: Guilford Press.

Keltner, D., & Ekman, P. (2000). Fa-cial expression of emotion. In M.Lewis & J.M. Haviland-Jones(Eds.),

Handbook of emotions

(2nded., pp. 236–249). New York: Guil-ford Press.

Messinger, D.S., Fogel, A., & Dick-son, K.L. (2001). (See References)

Acknowledgments—

This research wassupported by two James W. McLamoreAwards and by a grant from the NationalInstitute of Child Health and Human De-velopment (HD38336). I thank Alan Fo-gel , Harr ie t Oster , Laura Bolzani ,Cornelia Gro

�

, Jo-Anne Bachorowski,and Rod Wellens.

Notes

1. Address correspondence to Daniel

S. Messinger, Psychology Annex, P.O.Box 249229, Coral Gables, FL 33124-0721; e-mail: [email protected].

2. We created each pair by pastingidentical lower portions of an expres-sion (from the cheekbones down) overone expression with and one expressionwithout cheek raising. However, eventhe cry-faces classified as lacking cheek

ing particularly intense compositeexpressions. These expressionsmight be particularly salient tononexpert observers and so mightbe economical indices of tempera-ment and of emotion regulation inboth normally developing and at-risk infants.

New simulation software prom-ises to illuminate how facial ex-pressions are patterned in timewith other expressive actions suchas vocalizations and gazes at apartner. Studies using this softwareindicate that infants embed vocal-izations within the course of ongo-ing smiles and cry-faces (Yale,Messinger, Cobo-Lewis, Oller, &Eilers, 1999). Infants may be callingattention to their emotional expres-sions, as the facial expressions tendto begin during a gaze at mother.Infants’ expressive actions influ-ence and are influenced by similarexpressive actions on the part ofcaregivers. Using simulation soft-ware to investigate how infantsand caregivers coordinate their ac-tions in time will illuminate the in-terpersonal contexts that are simul-taneously a primary context inwhich infant emotion occurs andpart of an emotional process occur-ring between two partners.

What about development be-yond infancy? There appear to besimilarities between infants’ andadults’ smiles, though their smileswithout cheek raising appear tohave different functions. Adultsalso appear to be more inclinedthan infants to express negativeemotion discretely (e.g., using dif-ferent expressions for anger andsadness) rather than through cry-faces. How and when do such dif-ferences between infants and

raising involved a weak level of this ac-tion, so there was a bias against findingan effect of cheek raising. Mouth open-ing was defined as dropping of the jaw.Stronger smiles involved stronger lip-corner raising. Stronger cry-faces in-volved stronger sideways lip-cornerpulling and the presence of chin rais-ing. Anatomically based descriptions ofthe stimuli and details of the statisticalanalyses and procedure are availablefrom the author.

References

Camras, L.A. (1992). Expressive development and

basic emotion.

Cognition and Emotion

,

6

, 269–283.Cohn, J.F., Zlochower, A.J., Lien, J., & Kanade, T.

(1999). Automated face analysis by featurepoint tracking has high concurrent validity

with manual FACS coding.

Psychophysiology

,

36

, 35–43.Ekman, P. (1994). All emotions are basic. In P. Ek-

man & R.J. Davidson (Eds.),

The nature of emo-tion: Fundamental questions

(pp. 15–19). NewYork: Oxford University Press.

Ekman, P., Davidson, R.J., & Friesen, W. (1990).The Duchenne smile: Emotional expressionand brain physiology II.

Journal of Personalityand Social Psychology

,

58

, 342–353.Fogel, A., Nelson-Goens, G.C., Hsu, H., & Shapiro,

A.F. (2000). Do different infant smiles reflect dif-ferent emotions.

Social Development

,

9

, 497–520.Fox, N., & Davidson, R.J. (1988). Patterns of brain

electrical activity during facial signs of emo-tion in 10 month old infants.

DevelopmentalPsychology

,

24

, 230–236.Matias, R., & Cohn, J.F. (1993). Are MAX-specified

infant facial expressions during face-to-face in-teraction consistent with differential emotionstheory?

Developmental Psychology

,

29

, 524–531.Messinger, D.S., Fogel, A., & Dickson, K.L. (1999).

What’s in a smile?


,

35

, 701–708.Messinger, D.S., Fogel, A., & Dickson, K.L. (2001).

All smiles are positive, but some smiles aremore positive than others.

Developmental Psy-chology

,

37

, 642–653.Oster, H. (2000).

Baby FACS: Facial Action Coding Sys-tem for infants and young children

. Unpublishedmanuscript, New York University, New York.

Oster, H., Hegley, D., & Nagel, L. (1992). Adultjudgments and fine-grained analysis of infantfacial expressions: Testing the validity of a pri-ori coding formulas.


,

28

, 1115–1131.Yale, M., Messinger, D.S., Cobo-Lewis, A.B., Oller,

D.K., & Eilers, R.E. (1999). An event-basedanalysis of the coordination of early infant vo-calizations and facial actions.

DevelopmentalPsychology

,

35

, 505–513.

Documents

Positive and Negative: Infant Facial Expressions and … dominant theory holds that ... discrete emotions, such as joy and anger (Ekman, 1994). Each ... Positive and Negative: Infant