Symposium
Chair: Daniel S. Messinger
Different types of smiling have recently been documented in infancy, buttheir emotional significance is the subject of theoretical controversy. Someperspectives suggest infants experience an innate, unitary feeling of joywhile others suggest the possibility of multiple, linked positive emotions.The symposium addresses this issues with new research on the identificationand first appearance of different types of smiling, and their roles indifferent types of social interaction.Cohn, Tian, Forbes, Ambadar, and Kanade present data on a computer programbeing developed for the automatic recognition of videotaped, real-timefacial actions among infants and their caregivers. The program outputsfacial Action Units (AUs) described in Ekman and Friesen's Facial ActionCoding System (FACS). FACS codes are the basis for the distinctions betweendifferent types of smiling used by the symposium's participants: Simplesmiles involve only lip corner raising (AU12); in Duchenne smiles, thecheeks are raised high around the eyes (AU12+AU6); open mouth or play smilesinvolve the jaw dropping (AU12+AU26/27); duplay smiles result fromsimultaneous Duchenne and open mouth smiling (AU12+AU6+AU26/27). The programshows impressive agreement with human coders in distinguishing betweenDuchenne and non-Duchenne smiles, and in recognizing different degrees ofmouth opening. Dondi, Messinger, Colle, Beghi, Simion, & Fogel begin the developmentalstory by documenting bilateral Duchenne smiling in the newborn. The smilesare relatively frequent during states of active sleep and drowsiness andhave longer durations than other smiles. The authors address the possibilitythat these smiles have emotional significance by integrating neuroscientificfindings on the role of the amygdala in early emotional processing and motoractivity. Some emotion theorists have argued that the Duchenne smile is a unique indexof positive emotion. Messinger, Fogel, and Dickson find evidence thatbetween one and six months, Duchenne smiles are associated with a positiveperiod of social interaction, occurring when mothers are smiling.Independently, however, open mouth smiles also tended to occur when motherswere smiling and when infants were gazing at their mothers' faces, anotherpositive period of social interaction. They argue that both Duchenne andopen mouth smiling are associated with positive emotion in infancy andsuggest situational similarities between the different types of smiling.Fogel, Hsu, and Shapiro argue that different infant smiles reflect differenttypes of positive emotions among 6- and 12-month-olds. Their experimentalprotocol involved examining the distribution of different types of smiles inthe set-up and climax phases of peek-a-boo and tickle games, when infantswere gazing at and away from mother. Synthesizing this information, theyargue that simple, Duchenne, play and duplay smiles reflect different typesof positive emotion as indexed by the infant's involvement with and responseto the different phases of these ongoing games. Provocatively, they also askwhether the same type of smile might not have a different emotionalsignificance depending on whether it occurred, for example, in the set-upphase of a peek-a-boo game or the climax phase of tickling game.
Details of individual items:
paper
Facial displays are a primary means of communication for preverbalinfants and their caregivers. They communicate emotion (Izard & Malatesta,1987) and pain (Izard et al., 1983), regulate social behavior (Cohn &Tronick, 1983), indicate brain (Fox & Davidson, 1988) and autonomicactivation (Weinberg & Tronick, 1996) and pathology (Rinn, 1984), and signaldevelopmental transitions (Campos, Bennett, & Kermoian, 1992). Infants havewell-organized emotional responses to the facial displays of caregivers(Cohn & Elmore, 1988), and stable patterns of individual differences infacial expression arise within the first half year (Cohn & Campbell, 1992).Individual differences in facial expression in response to age-appropriatestressors predict later social and emotional development (Cohn et al.,1992). Because of the importance of infant and caregiver facial displays ininfancy research, reliable, valid, and efficient methods of measurement arecritical. With extensive training, human observers can achieve acceptablelevels of inter-observer reliability in coding facial displays.Human-observer-based (i.e., manual) methods, however, are labor intensive,semi-quantitative, and difficult to standardize across laboratories or overtime. Training is time consuming, and coding criteria may drift with time(Martin & Bateson, 1986). Such extensive effort discourages standardizedmeasurement and may encourage the use of less specific coding systems withunknown convergent validity (Matias et al., 1989). These problems tend topromote the use of smaller sample sizes (of subjects and behavior samples),prolong study completion times, and thus limit the generalizability of studyfindings. To efficiently capture the full range of infant and caregiver facialexpression, automated detection, tracking, and classification offine-grained changes in facial features are needed. Our inter-disciplinarygroup of psychologists and computer scientists has developed the firstversion of a computer vision system that is sensitive to subtle changes inthe face. The system includes three modules to extract feature information:dense-flow extraction using a wavelet motion model, facial feature tracking,and edge and line extraction. Dense-flow refers to the direction andmagnitude of pixel motion between successive video images. The featureinformation thus extracted is fed to discriminant classifiers or artificialneural networks that classify it into FACS action units [Cohn et al., 1999;Lien et al., in press; Tian et al., in press], the descriptive system tocode fine-grained changes in facial expression. In previous work, thesystem was tested on image sequences from male and female adult subjects ofvaried (Cohn et al., 1999; Lien et al. in press). In this presentation, wereport results of recent and on-going validity studies in infants and theircaregivers. Infants experienced procedures designed to elicit a wide-range ofemotion expressions: a stranger- and mother-approach, maternal separation,growling gorilla task, and mother-infant face-to-face interaction. Thesetasks elicited expressions of fear, anger, and joy. The mother-infantinteraction provided an opportunity to quantify fine-grained changes inmother-infant synchrony. Results of face detection, feature tracking, andaction unit recognition will be presented for each context. The methods,although still in development, afford a powerful and efficient method ofquantifying the temporal organization of emotion displays and parent-infantsynchrony.
paper
Recent evidence indicates that a certain type of smile, the Duchenne smileinvolving high cheek raising, is especially associated with positive emotionin adults and older infants. Despite anecdotal reports, there have been noquantitative studies of Duchenne smiles in the period immediatly afterbirth. Unlike the 'social smiling' described by other authors, the neonatalendogenous smiling is not the result of external stimulation. However,little is known about the significance or developmental relevance for latersmiling of this spontaneous behavior. Like other motor patterns, neonatal smiling is considered a behavior whichoccur in the absence of known external or internal (visceral) stimuli. Werecently conducted a series of studies of the prevalence, frequency, andduration of different types of neonatal smiles and their relationship tobehavioral state. 32 full-term, healthy neonates (X22,8 hrs postnatal) werevideotaped during six minutes of quiet sleep, active sleep, drowsiness andalertness. States were defined in terms of conjunctions of behavioralconditions, incorporating salient features of several state coding systems,currently in use. Smiles were identified and analyzed using anatomicallybased coding systems (Baby FACS: Rosenstein & Oster, in press). Analyses ofthe data were carried out on two measures: the frequency, and the meanduration of each non-Duchenne smile (AU12), and each Duchenne smile(AU12/AU6). Bilateral Duchenne smiles occurred at a rate of 1,9 per 6 minutes, A ratethat was not significantly different from that of bilateral non-Duchennesmile, 2,6 per 6 minutes. Though these Duchenne smiles were brief (X1,87secs.), they had longer durations that the comparable bilateralnon-duchenne smiles (X1,38 secs., p<.005). Smiling occurred predominantlyduring the states of active sleep and drowsiness, rather than during quietsleep and alertness. The finding that neonates can produce Duchenne smiles, tell us that thisfacial expression is part of their innate repertoire, but it does not tellus how they function in the neonate's psychological economy. Mere productionof a Duchenne smile, in fact, does not necessarily indicate the experienceof enjoyment. On the other hand, many authors suggested that the amount andvariability of autonomic activity seen during active sleep is highlysuggestive of limbic activation. The amygdaloid complexes, for example,receive an important set of afferent nerve connections arising from thedorsal pons and intralaminar thalamic nuclei, by which they could beactivated during sleep. In addition, the amygdaloid complexes projects backto subcortical structures (hypothalamus and several brainstem nuclei). Bythese subcortical projections, the amygdaloid complexes could induce theautonomic manifestations that are commonly observed during REM sleep. Giventhe intimate connectivity between the amygdala and the brainstem systeminvolved in the control of preprogrammed species-typical behavior related tothe emotional response, emerging data from neurobiology suggest toreconsider early explanations of neonatal smile, recommending further andmore refined investigations.
paper
Positive emotion such as joy is thought to motivate infant action,> signaling to self and other that an activity should be continued. Certain> types of smiling have been hypothesized to be uniquely associated with> positive emotion. However, data linking specific types of infant smiling> to emotion eliciting situations are sparse. The current study tests the> hypothesis that smiles hypothesized to be indices of positive emotion are> more likely than other smiles during periods of early social interaction> thought to elicit positive emotion. > 13 infants were observed weekly from one to six months of age in> face-to-face interactions with their mothers. Smiling (AU12), high cheek> raising (AU6) and jaw dropping (AU26C-AU27) were coded reliably (by FACS> certified coders trained in Baby FACS). This coding was used to identify> periods of Duchenne and non-Duchenne smiling (smiling with and without> high cheek raising, respectively) and open and closed mouth smiling> (smiling with and without jaw dropping, respectively). Periods of gazing> at mother's face and mother smiling were also coded reliably. We thought> these positive periods of interaction might be differentially associated> with Duchenne and open mouth smiling. Loglinear analyses with follow-up> t-tests identified a set of independent, non-overlapping associations> between type of smile and period of interaction. Only when mother smiling> preceded infant smiling were co-occurrences involving mother smiling> included in the analyses. > Duchenne smiling was more likely to occur during periods of mother smiling> than was non-Duchenne smiling (p < .05). However, Duchenne smiling was not> more likely to occur during periods of gazing at mother's face than> non-Duchenne smiling (p .64). Open mouth smiling was more likely to> occur both during periods of mother smiling and periods of gazing at> mother's face than was closed mouth smiling (ps < .025 & .01,> respectively). > The results indicate that smiles hypothesized to be particularly joyful> are more likely than other smiles to occur during positive periods of> social interaction. This is a challenge to the perspective that all> smiling indexes a single positive emotion. The results also challenge the> possibility that Duchenne smiling is a unique index of positive emotion in> infancy. Infant Duchenne smiling had unique associations with one positive> period of social interaction, mother smiling. Open mouth smiling had> unique associations with two positive periods, mother smiling and gazing> at mother's face. A conservative interpretation is that, in early infancy,> open mouth smiling as well as Duchenne smiling indexes positive emotion. > Although only unique associations between smiling and positive periods of> interaction were examined, previous reports from this data set indicate> that over half of Duchenne smiling involved mouth opening, and vice-versa.> The temporal overlap of these smiles and the independent associations both> have with mother smiling suggest Duchenne and open mouth smiling may be> part of a related family of joyful emotions.
paper
Smiling is a facial action that is considered to be the prototypical> expression of pleasure. There are, however, different types of smiles that> occur in a wide variety of social contexts in adults. Smiles differ> according to whether lip corner retraction is accompanied by different> facial movements such as the presence or absence of eye wrinkling or mouth> opening. The recent discovery of different types of smiles in young> infants has created renewed interest in positive emotional development> which has been studied less than distress and coping. These different> smiles raise important questions about emotions and their development. Are> there different types of positive emotion in early infancy? Do the> different expressions reflect variations on the same emotion? Can each> expression reflect different emotions depending upon the context? > Different types of infant smile in the family of positive emotions> were investigated during two mother-infant games: peekaboo and tickle.> Both games have an invariant structure involving setup and climax> components. For peekaboo, setup involves the covering or disappearance of> the mother's face ('Where's mommy(?)'), and climax involves its uncovering> or reappearance ('Peekaboo'). For tickle, setup involves an approach of> the mother's hands toward the infant's body ('I'm gonna get you'), and> climax involves the act of tickling (which may be accompanied by maternal> vocalization). Both games require vision in the setup while in tickle the> modality switches from visual to tactile. The similar invariant structure> of the two games makes it relatively easy to compare smiling across> components and between games. There were 27 6-month-old infants and 28> 12-month-olds. Infant smiles were coded as simple (lip corner retraction> only), Duchenne (simple plus cheek raising), play (simple plus jaw drop),> and duplay (simple plus cheek raise and jaw drop). Infant gaze direction> was coded as either at mother or away.> Results show that each type of smile has a systematic pattern of> association with the game (peekaboo or tickle), component (setup or> climax), trial (six trials for each game), and the direction of the> infant's gaze. No significant age differences were found. As suggested by> the social process theory of emotion, the emotional significance of> different types of smiling depends upon the action readiness implied by> the smile (recruitment of the eyes and/or the mouth), the infants's gaze> direction (indexing the infant's attention to visual compared to tactile> events), and the location of infant action within the sequential social> process of the game (in setup or climax, or in early or later trials). The> different smiles seemed to be related to uniquely different emotions and> that each type of smile may have more than one emotional meaning depending> upon the infant's gaze direction and the sequence of other actions. Our> results, therefore, suggest that there are multiple forms of enjoyment> even within the simple context of these two different mother-infant games.> Emotional meaning, therefore, is not indexed by particular expressions but> by types of readiness to engage in communicative action, types that are> regulated in part by the temporal dynamics of how social processes> naturally unfold in time.