Abstract
Emotion theories based on research with adults must be able to accommodate developmental data if they are to be deemed satisfactory accounts of human emotion. Inspired in part by theory and research on adult emotion, developmentalists have investigated emotion-related processes including affect elicitation, internal and overtly observable emotion responding, emotion regulation, and understanding emotion in others. Many developmental studies parallel investigations conducted with adults. In this article, we review current theories of emotional development as well as research related to the several aspects of emotion designated above. Beyond providing an overview of the field, we hope to encourage greater cross-fertilization and research collaboration between developmental psychologists and scholars who focus on adult emotion.
Keywords
The study of emotional development languished for decades during the mid-20th century as scholars focused their attention on cognitive development under the influence of Piaget’s landmark theory. Nevertheless, inspired by the revival of interest in adult emotion, some developmentalists began to turn their attention to infant and children’s affect starting in the 1970s. Like their adult-oriented colleagues, developmental researchers subscribe to the notion that emotion-related processes include affect elicitation, internal and overt observable responses, emotion regulation, and understanding emotion in others. These aspects of emotion are inextricably interrelated and thus, the topical divisions we use in this review are somewhat artificial. Nonetheless, this should not impede our goal of presenting an overview of current theory and research.
Theories of Emotional Development
In 1932, Katherine Bridges published a highly influential monograph, which dominated the literature on emotional development for several decades. Bridges described infant emotions as originating in a state of diffuse excitement that first differentiates to generate delight and distress and then more distinct emotion states such as fear, anger, elation, and affection. Although Bridges’s account was primarily descriptive in nature, she implicitly attributed the timing of emotional development to maturational processes.
More recently, several theories of emotional development have been proposed that align themselves with competing contemporary views of adult emotion. According to Carroll Izard’s differential emotions theory (DET; Izard & Malatesta, 1987), discrete basic emotions also emerge according to a maturational timetable. However, in contrast to K. M. B. Bridges (1932), DET asserts that discrete basic emotions (e.g., anger, fear) are not preceded by less differentiated affective states (e.g., distress). Instead, the core constituents of emotion (currently conceived as neural and experiential; Izard, 2009) are hard-wired systems with distinct motivational properties that emerge de novo at specific preprogrammed ages. After the emergence of an emotion, its further development involves elaboration via integration into a cognitive affective “scheme” that involves links between the emotion itself and associated eliciting situations, instrumental actions, and regulatory responses.
In contrast to DET, Campos’s functionalist view of emotional development does not assume the existence of hard-wired emotion systems involving invariant neural or even experiential core components. Instead an emotion is a relational process through which an individual attempts to establish, change, or maintain some significant aspect of his or her relationship to the external or internal environment (K. Barrett & Campos, 1987; Campos, Walle, Dahl, & Main, 2011). Thus, emotional responses are wholly functional rather than fixed or preprogrammed.
Functionalists have delineated a useful list of phenomena that are involved in emotional development (J. J. Campos, Barrett, Lamb, Goldsmith, & Stenberg, 1983). These include changes in: (a) relational goals, (b) interpretation of the environment in relation to those goals, (c) capacities that may enable the pursuit of emotional goals (e.g., language, motoric capabilities, cognitive functioning), and (d) capacities related to the ability to revise (i.e., regulate) initial emotion response tendencies. For example, anger may be conceptualized as an emotional process in which the individual attempts to overcome an impediment to a goal that is perceived to be of significance to the individual. During childhood, some children may develop the goal of preserving their reputation with peers and a child may then respond with anger to indications of disrespect by other children. However, interpretation of peer behaviors as indications of disrespect also undergoes development as children learn that certain verbal declarations are intended as insults (e.g., “Mama’s boy!”). Responses to such reputational impediments also change. For example, as children become more linguistically adept, they may respond with counterinsults rather than physical aggression. In addition, children may learn to defer producing an immediate anger response in the service of a competing relational goal (e.g., to avoid getting into trouble). However, developments in cognitive skills (e.g., memory, executive functioning) may eventually enable children to exact their revenge at a later time.
One emerging contemporary perspective on emotional development is the dynamical systems approach (see Camras & Witherington, 2005, for review). Like the functionalist perspective, the dynamical systems approach emphasizes flexibility rather than fixity in emotion responding. With some exceptions, proponents of both views assert that any specific emotion episode may or may not include a particular emotion component (e.g., an emotional facial expression). Such variability may reflect top–down cognitive processes (e.g., deciding to defer expressing one’s anger). However, dynamical systems theorists place greater emphasis on the influence of bottom–up contextual factors that may or may not be cognitive in nature. For example, Camras and her colleagues have shown that facial expression production is sometimes determined by nonemotion factors such as gaze direction or co-occurring instrumental behaviors. To illustrate, Camras, Lambrecht, and Michel (1996) found that 5- to 7-month-old infants produce “surprise” expressions when opening their mouths to incorporate a familiar object (e.g., put a familiar toy in their mouth).
Dynamical systems proponents posit qualitative shifts in the organization of emotions (termed “phase shifts”) that may occur both in real time and also during the course of development. Such shifts will occur when some key component of the system reaches a critical threshold. For example, in real time, infants may shift from smiling to crying when the intensity of stimulation reaches a certain level. Across development, major shifts in emotion responding may occur when children reach critical developmental milestones in domains such as cognitive, language, and motor development (e.g., self-produced locomotion via crawling; J. Campos, Kermoian, & Zumbahlen, 1992).
Interestingly, some developmentalists have produced theories that include elements of several perspectives described before. For example, Alan Sroufe’s (1996) theory of emotional development retains elements of earlier differentiation theories (i.e., discrete emotions deriving from less differentiated earlier reactions). However, like dynamical systems proponents, he describes emotional development in infancy as involving a series of qualitative shifts that he ties to cognitive development. Holodynski and Friedlmeier (2006) have proposed an internalization model that also involves emotion differentiation, but they attribute such differentiation more to sociocultural shaping. Camras (2011) has proposed a framework incorporating contributions from differentiation, dynamical systems, and functionalist perspectives. She views emotions as dynamical systems attractor states that also become more differentiated with age, but attributes developmental changes in emotion to a broader range of factors including (but going beyond) those identified within the functionalist framework.
Emotion Elicitation
A variety of processes may lead to emotion elicitation. For example, emotions may be induced as the result of conditioning (e.g., Watson’s famous example of Little Albert) or conscious appraisal of experiences in relation to one’s goals (e.g., losing a favorite toy). However, developmentalists have primarily approached the question of emotion elicitation from two perspectives: temperament and socialization.
Temperament
Although several models of temperament have been proposed (see Dyson, Olino, Durbin, Goldsmith, & Klein, 2012), that of Rothbart and her colleagues is currently generating the most active research. Rothbart, Sheese, Rueda, and Posner (2011) define temperament as “constitutionally based individual differences in reactivity and regulation in the domains of affect, activity, and attention” (p. 207). Based primarily on parent reports using questionnaire measures, several categories of affective reactivity are included in Rothbart’s model: fearful distress/fear, frustration/irritability/anger, sadness, and positive affectivity (Rothbart & Bates, 2008). Individual differences in the manifestation of these emotions are due to both differences in reactivity and also to differences in the regulation of these reactive tendencies. While temperament is thought to have an inherent basis (e.g., related to genetic variation; Rothbart & Bates, 2008), the role of environment is recognized in that characteristic response tendencies may both be relatively stable yet change over time as a result of experience. For example, in a short-term longitudinal study, Calkins (2002) found continuity in the level of anger/frustration shown by infants who experienced negative parenting, but not for infants who experienced positive parenting. Similarly, in an extensive program of longitudinal research on behavioral inhibition, Kagan and his colleagues (see Kagan & Fox, 2006) have shown that many children categorized as highly inhibited in early childhood (i.e., who presumably would score high on fear and shyness using Rothbart’s measures) become substantially less inhibited later in development. Nevertheless, early temperament can be a predictor of later socioemotional adjustment as has been demonstrated in several studies. For example, in a widely cited longitudinal investigation, Caspi and Silva (1995) found that “undercontrolled” 3-year-old children (who showed more anger/frustration and lower attentional control) were more likely to later report less satisfactory social and emotional relationships and more antisocial behavior during adolescence and adulthood.
Emotion Socialization
While nonsocial experiences may be important influences on emotion elicitation (e.g., falling may produce fear of heights), most developmental research focuses on the actions and activities of social agents. We begin by presenting a conceptual model of parents’ emotion socialization proposed by Eisenberg and her colleagues (Eisenberg, Cumberland, & Spinrad, 1998; Eisenberg, Spinrad, & Cumberland, 1998), although other excellent models have been proposed that have much in common with that of Eisenberg et al. (e.g., Halberstadt, Denham, & Dunsmore, 2001; Holodynski & Friedlmeier, 2006; Saarni, 1999). While focusing on parents’ behaviors with respect to children, the model can be readily extended to cover other socialization agents as well as infants and adolescents as socialization targets.
Eisenberg and colleagues (Eisenberg, Cumberland, & Spinrad, 1998; Eisenberg, Spinrad, Cumberland, 1998) distinguish four categories of parental emotion-related socialization behaviors (ERSBs): (a) parents’ own emotional expression, (b) reactions to the child’s emotions (i.e., contingent responding), (c) discussion of emotion with or in the presence of the child, and (d) management of children’s exposure to emotion-inducing situations. Importantly, the effects of parental behaviors on children’s emotions may be moderated by a number of factors (e.g., type of emotion involved, temperament of the child, whether parental behavior is directed toward the child or toward another). Cultural variability also characterizes emotion socialization (see Cole & Tan, 2007, for review).
Parents’ own emotional expression is an important source of knowledge about emotion. For example, once infants begin to engage in social referencing (i.e., seeking information about others’ emotional responses when they are unsure how to respond themselves; J. J. Campos & Stenberg, 1981), they begin to learn what types of stimuli should evoke particular emotions. Similarly, parents’ contingent responses to children’s spontaneously expressed emotion provide information regarding whether the parent considers the child’s emotion to be appropriate. Parents may reinforce, ignore, punish, and/or model an alternative emotional response to that displayed by the child. For example, in a classic study of family conflict, Patterson (1982) found that parents of aggressive children tend to both model and reinforce anger reactions to perceived provocations. Discussion of emotion also may occur during family conflicts, for example, the child may be instructed regarding what emotion should properly be experienced. However, emotion-socializing discussion also takes place at other times (e.g., while observing episodes of other children’s emotion, reading stories that involve emotion, recalling shared past experiences). For example, Miller, Fung, Lin, Chen, and Boldt (2012) found that when relating stories involving children’s past transgressions, Taiwanese mothers often indicated that the child should feel shame. Lastly, regarding Eisenberg’s fourth category, parents may introduce the child to emotionally challenging situations in order to provide opportunities for emotion socialization. Combining examples of the several emotion-socializing behaviors in a single episode, parents may respond to their child’s fear of dogs by introducing the child to a friendly puppy while demonstrating positive affect and explaining to the child why dogs should not to be feared.
Emotion Responding
While elicitation refers to the “when” component of the emotion process, emotion responding refers to the “how,” that is, how emotions are manifested in behavioral output or internal response processes. Perhaps the biggest challenge to studying emotional development is the fact that there is no “gold standard” of observable behavior that can be used to determine whether or when an emotion has been produced (J. Campos, Frankel, & Camras, 2004). This is particularly problematic for studies of infants and young children who cannot verbally report their emotion experience. Differential emotions theory originally attempted to solve this problem by positing a unique inherent link between specific facial expressions and their corresponding discrete emotions (Izard & Malatesta, 1987). However, evidence for such a strict relationship is lacking even in infancy (Camras & Fatani, 2008; Camras & Shutter, 2010). In consequence, evidence that a particular response reflects a particular emotion in infants and young children must rest on observing systematic emotion-appropriate correspondences between eliciting circumstances, facial and vocal expressive behaviors, and nonfacial responses. In the following discussion, we will consider these three components and their interrelationships.
Expressive Behavior and Action Tendencies
Facial and vocal expressions are present from birth and have received the most attention in studies of infancy. Neonatal crying is acknowledged to indicate distress as it generally occurs in situations that would be expected to elicit a distress response (e.g., inoculation, hunger; R. G. Campos, 1989). In contrast, neonatal smiling is more ambiguous. Neonatal smiling occurs in response to gentle sensory stimulation (e.g., stroking the forehead) and may be seen during sleep (Emde & Harmon, 1972). Some neonatal smiles include the cheek raising component proposed to distinguish genuine versus nongenuine expressions of happiness (Messinger et al., 2002), but some neonatal smiles do not. Irrespective, most developmentalists follow Sroufe (1996) in interpreting newborns’ smiles as reflecting the reduction of tension engendered by mild stimulation, rather than as an expression of happiness.
Starting at around 6 weeks of age, infants begin to produce smiles in social situations, particularly in response to their caregivers (Emde & Harmon, 1972). Such “social smiles” are generally interpreted as reflecting happiness/joy (primarily because they occur in situations during which positive emotion would be expected to occur). “Mastery” smiles also begin to occur in situations that would be expected to induce positive emotion (e.g., achieving a goal after some effort; Lewis, Alessandri, & Sullivan, 1990). However, by 3 years of age, some children also smile in negative emotion situations. For example, they will smile to the gift giver after receiving a disappointing present, although in private they express more negative emotion (Cole, Zahn-Waxler, & Smith, 1994). Such findings suggest that at least some preschool children can dissimulate their expressive behavior.
While there is general agreement about which facial components indicate negative affect in infants (e.g., contracting the brows, lowering the lip corners), there is considerable controversy as to whether different negative expressions reflect different discrete negative emotions (e.g., anger vs. fear vs. sadness as opposed to nonspecific distress; Camras & Shutter, 2010). As indicated earlier, while DET describes distinct facial expressions for each discrete negative emotion, there is little evidence that during the first year these expressions are differentially distributed across situations in which different negative emotions are being induced. For example, Camras et al. (2007) found that 11-month-old Chinese, Japanese, and American infants produced similar facial expressions in response to both nonpainful arm restraint and presentation of a disembodied toy gorilla head.
Interestingly, although facial expressions were not differentiated in the Camras et al. (2007) study, raters viewing videotapes of the infants (with facial expressions electronically obscured) judged the babies to be experiencing different negative emotions (primarily fear during the gorilla presentation and anger/frustration during arm restraint). Thus raters could use situational and nonfacial cues to make judgments about infants’ discrete negative emotions. Objective coding of the infants’ nonfacial behaviors determined that infants showed more behavioral stilling during growling gorilla and more physical resistance during arm restraint. Presumably observing these behaviors influenced the raters’ emotion judgments. Nonetheless, these behavioral responses also are not unique to fear or anger. For example, stilling may occur in surprise and interest situations as well as during episodes of fear.
Still, some adult-oriented theories of emotion propose that different action tendencies are inherently linked to different discrete emotions (e.g., Frijda, 1986). While theoretically plausible, research with even younger infants has not produced clear-cut evidence for this position. Camras, Sullivan, and Michel (1993) found that limb movements accompanying DET-designated “anger” expressions in a 4- to 9-week old infant were more active and jerky than those accompanying DET-designated “sad” expressions, and that “sad” expressions were accompanied by little movement overall. However, no goal direction in the infant’s movements could be discerned.
Using a contingency-learning procedure in which infants can produce a desirable audiovisual stimulus by moving their arm, Lewis et al. (1990) found concordance between increased arm pulling and increased production of DET-predicted “anger” expressions during extinction. However, Sullivan and Lewis (2003) subsequently found an inverse relationship (i.e., decreased arm pulling and increases in “anger” expression) when contingent learning was followed by loss of the contingency (i.e., losing control of stimulus presentation). Thus, covariance between facial and behavioral responding disappeared when arm pulling was discovered to no longer be functional. These results are consistent with the functionalist principle of equipotentiality (J. J. Campos et al., 2011), indicating the arm pulling and facial expressions operate independently according to their situationally specific functionality even during infancy.
During the second year, infants’ motoric and cognitive capabilities increase and they begin to show situation-related behaviors that lead many researchers to infer the presence of new emotions. For example, when made the center of attention, toddlers often show signs of avoidance (e.g., lowering their heads, avoiding eye contact and/or hiding their face; Lewis, 1995). These situationally embedded responses are interpreted as indicating embarrassment. After accidentally breaking a toy, 2-year-olds may hide their transgression and avoid the experimenter or may approach the adult and attempt to repair the damage. These situationally embedded responses have been interpreted as indicating shame (in the first case) or guilt (in the latter case; K. C. Barrett, Zahn-Waxler, & Cole, 1993). Thus shame and guilt are not manifested through distinct facial expressions, but rather through behaviors that appear to facilitate the functional goals associated with these “self-conscious” emotions.
As infants grow older, development of more advanced means–ends behaviors and executive functioning allows children to employ a wider range of instrumental behaviors in the service of emotion-related goals (J. J. Campos et al., 2011; Thompson, 2011). This is adaptive since actions required to achieve one’s goals vary greatly across situations. As we will argue further in what follows, goal-directed emotion-related actions can be conceptualized as being forms of “regulated” emotional behaviors.
Neurophysiological Responses
In contrast to many researchers studying adult emotion, developmentalists have not been concerned with identifying distinct neural substrates for different discrete emotions. Instead, most research attempts to identify indicators associated with well-being and maladjustment (Dennis, Buss, & Hastings, 2012). Four types of measures are commonly used: electroencephalography (EEG) measures (e.g., of cerebral lateralization), heart rate measures (e.g., vagal tone and vagal suppression), functional magnetic resonance imaging (fMRI; e.g., amygdala activation) and neurohormonal measures (e.g., cortisol responding, oxytocin). Many findings parallel those obtained in similar studies with adults. For example, Davidson and Fox (1989) demonstrated that infants and children who show relatively greater left frontal EEG activation are also more likely to show positive affect and approach behavior, while those who show greater right frontal activation show more negative affect and behavioral inhibition. Other studies have found EEG differences between children with different types of behavior problems (Stieben et al., 2007). Regarding heart rate measures, low vagal tone (i.e., high resting heart rate and low variability) has been associated with adverse outcomes including greater behavioral inhibition and negative emotionality (Beauchaine, 2001; Hinnant & El-Sheikh, 2009; Kagan & Fox, 2006). While relatively few fMRI studies have been conducted with children, their results have been largely consistent with the adult literature. For example, Thomas et al. (2001) found that anxious children showed increased amygdala activation in response to negative stimuli while depressed children showed a lesser response. Cortisol reactivity has been studied primarily as a stress response and has been linked to emotional difficulties and psychopathology in children (Gunnar & Vasquez, 2006; Heim, Owens, Plotsky, & Nemeroff, 1997; Strang, Hanson, & Pollak, 2012). Research on cortisol reactivity has also highlighted the influence of experience on physiological responding. For example, infants and young children raised in adverse circumstances (e.g., maltreating families, international orphanages) often demonstrate atypical patterns of cortisol reactivity (Cicchetti & Rogosch, 2012; Gunnar, Morison, Chisholm, & Schuder, 2001).
Emotion Responding and Emotion Regulation
The conceptual distinction between emotion and emotion regulation has been widely debated in the developmental literature as well as the nondevelopmental literature (e.g., Cole, Martin, & Dennis, 2004; Gross & Feldman Barrett, 2011). Tied to this debate are methodological challenges involved in attempting to distinguish regulated from unregulated emotion behaviors. In their widely cited article, Cole et al. (2004) presented a number of suggestions regarding methodologies for distinguishing spontaneous emotion responding from emotion regulation. However, an alternative would be to avoid making distinctions between spontaneous/unregulated versus regulated emotion and consider all forms of emotion responding to be the product of multiple influences that include both biological predispositions and past experiences (J. Campos et al., 2004; J. J. Campos et al., 2011; Camras, 2011; Thompson, 2011). Temporal changes in the form of one’s emotional behaviors over the course of an episode (e.g., “bringing one’s temper under control”) would be considered a shift or reorganization of emotional responding that occurs because new influencing factors (including emotion-related goals) are brought to bear on the situation (e.g., the realization that an important relationship is being jeopardized and needs to be preserved).
In early infancy, most emotion regulation/reorganization is actually induced by adults who may try to comfort or distract distressed infants (Kopp, 1989). However, the infant’s ability to shift attention is thought to influence the effectiveness of adult efforts and later to allow infants to sometimes regulate/reorganize their own emotion by disengaging from the distress-producing stimulus (e.g., looking away from a stony-faced mother; Cohn & Tronick, 1983). Individual differences in infants’ ability to shift attention are related to differences in emotional negativity (Rothbart et al., 2011). During the course of development, children gradually assume a greater role in organizing and reorganizing their own emotional behaviors (see L. J. Bridges & Grolnick, 1995). For example, they are less likely to seek help from their parents in an emotionally challenging situation (e.g., a delay of gratification task). Improvements in effortful control (i.e., children’s general ability to regulate their behavior) make a critical contribution to their ability to organize and reorganize their emotion responses. In addition, socialization plays an important role in influencing the behaviors that children produce in an emotion situation.
Socialization of Emotional Behavior and Regulation
The parenting behaviors described in Eisenberg et al.’s (Eisenberg, Cumberland, & Spinrad, 1998; Eisenberg, Spinrad, & Cumberland, 1998) model provide children with information about how (or how much) to behaviorally respond when an emotion is elicited. For example, parents’ positive expressivity is related to their children’s tendency to express their own positive emotion (Halberstadt & Eaton, 2002). Children’s positive expressivity is related to their greater social competence (e.g., lesser aggression) and better emotional adjustment (e.g., higher self-esteem; Halberstadt, Crisp, & Eaton, 1999). For negative emotion, however, the picture is different. That is, parent negative expressivity does not always (or only) result in imitation by the child. For example, the relationship between parents’ and children’s negative expressivity declines between toddlerhood and early childhood (although it increases again in late adolescence; Halberstadt & Eaton, 2002). As emphasized by social cognitive theory (Bandura, 1989), children’s imitative tendencies are moderated by many factors including the child’s interpretation of and own response to others’ behaviors. For example, young children may be frightened by parents’ anger and aggression and they may react by developing anxiety rather than (or in addition to) becoming bullies (Crockenberg & Langrock, 2001).
Another important factor moderating the relationship between parents’ and children’s emotional expression may be whether the parents’ expression is a contingent response to the child’s own emotional display. For example, to explain the declining relationship between parents’ and children’s negative expressivity during childhood, Halberstadt and Eaton (2002) speculated that parental anger in response to the child’s expression of negative affect may come to suppress the child’s own expression of negative emotion in the home. Further research on contingent responding suggests that disparaging or punishing a child’s undesirable emotional expression is less effective in encouraging socially desirable responding than is “emotion coaching” (i.e., acknowledging the child’s emotion and helping him or her learn to more effectively cope with the eliciting situation and their own emotional response; Fabes, Leonard, Kupanoff, & Martin, 2001; Gottman, Katz, & Hooven, 1997). For example, Morris, Silk, Steinberg, Myers, and Robinson (2007) found that mothers’ encouragement of attention refocusing and cognitive reframing were associated with the lessening of children’s angry and sad expressions after receipt of a disappointing gift. In contrast, Lunkenheimer, Shields, and Cortina (2007) found that disparaging or dismissing children’s emotions can be a risk factor for both externalizing and internalizing behavior problems in middle childhood.
Emotion coaching as a contingent response to children’s undesirable affect also involves Eisenberg, Cumberland, & Spinrad (1998) third category of emotion-related socialization behaviors, that is, discussion of emotion, However, as indicated earlier, emotion-socializing discussions also occur in situations during which the child is not currently experiencing the emotion (e.g., story-reading, observation of others’ emotion, recollections of past emotion episodes; Thompson, 2006). Numerous studies have found relationships between family discussion of emotion and children’s emotion understanding (e.g., Dunn & Brown, 1994).
Emotion Understanding
An important component of social and emotional competence is the ability to understand others’ emotions so that one may respond appropriately (Denham, 1998; Halberstadt et al., 2001; Saarni, 1999). We consider emotion understanding to be the ability to understand the relations among components of the emotion process, for example, eliciting situations and appraisals thereof, relational goals associated with each emotion, expressive and behavioral responses, verbal labels, and (arguably) internal feelings. For example, understanding the links between emotional facial expressions and the relational goals of the expresser (e.g., between disgust expressions and avoiding contact) would ideally allow children to predict the expresser’s behavior and generate a suitable response. Numerous studies have found that greater emotion understanding in children is related to greater social competence and emotional adjustment (e.g., Izard et al., 2011).
The capacity to understand emotions develops across infancy and childhood and is preceded by more basic abilities to discriminate among emotional expressions and match expressive stimuli across different modalities (e.g., happy faces with happy voices). Emotion discrimination is the ability to perceive morphological differences between emotional and nonemotional expressions as well as between different types of emotional expressions. Newborns demonstrate such differential responding to happy speech versus other emotional and nonemotional speech patterns (Mastropieri & Turkewitz, 1999). Four-month-old infants can discriminate among happy, angry, and neutral facial expressions (Serrano, Iglesias, & Loeches, 1995). By 5 months, infants smile more at positive versus negative infant-directed speech (Fernald, 1993) and show a preference for positive over negative facial expressions during face-to-face interactions (D’Entremont & Muir, 1999). At around 5 months, infants also demonstrate an ability to match emotional stimuli of different modalities based on valence (e.g., positive vocalizations with positive facial expressions; Walker-Andrews, 1986). According to Walker-Andrews (2008), bimodal (i.e., facial and vocal) processing facilitates the further development of emotion understanding. However, while 5-month-olds can discriminate and match expressive stimuli, there is still no clear evidence that these infants understand relationships between these expressive behaviors and other emotion components (e.g., elicitors and behavioral responses).
The first convincing evidence of such emotion understanding is older infants’ ability to seek and appropriately use others’ emotion signals to guide their behavior, as seen in social referencing studies (J. J. Campos, Thein, & Owen, 2003). For example, 8.5-month-old infants are more hesitant to approach a toy when their mother produces a negative rather than a positive vocalization. Similarly, 12-month-olds are more likely to cross a visual cliff when their caregiver displays a positive rather than a negative facial expression. By 12 months, infants can also use cues such as gaze direction and body posture to determine if an affective display is intended for a specific target. For example, Mumme and Fernald (2003) found that when infants viewed an experimenter produce a negative expression while looking at one of two objects, the infants subsequently avoided only the visually targeted object. By 10 months, infants also engage in affect sharing (e.g., by smiling and gazing back and forth between an object and the caregiver; Jones, Collins, & Hong, 1991; Venezia, Messinger, Thorp, & Mundy, 2004).
Although infants respond differentially to positive versus negative expression in social referencing studies, these investigations provide no strong evidence that they can distinguish among different emotions within the same valence (e.g., anger vs. fear). Beyond 14 months infants begin to use verbal emotion labels for different negative emotions in natural settings (Bretherton, Fritz, Zahn-Waxler, & Ridgeway, 1986). However, using more controlled experimental procedures, Widen and Russell (2008a, 2008b) concluded that evidence for differential understanding of negative emotions by 2-year-olds and even 3-year-olds is still not strong. Widen and Russell propose a differentiation model in which children first understand emotions only in terms of valence and arousal. Thus, preschoolers first differentiate among negative emotions according to how arousing they are. For example, when asked to choose facial expressions to put in a box marked “anger,” preschoolers first exclude expressions of low-arousal emotions such as sadness, followed by fearful, and finally disgust expressions (Russell & Widen, 2002). Differentiation continues throughout childhood and may proceed more rapidly for some emotions and emotion components than for others. For example, in emotion-naming tasks, children may perform better when presented with stories depicting causes and consequences of different negative emotions than with they do when presented with facial expressions (Widen & Russell, 2010). In contrast, they may perform better when presented with surprise facial expressions than with stories of surprise situations. Nonetheless, by the age of 5, children can generally match facial expressions and verbal labels to stories describing eliciting situations for a range of negative emotions (Camras & Allison, 1985). However performance on more complex tasks requiring the child to generate or respond to some verbal emotion labels and facial expressions (especially disgust expressions or the word “disgust”) may be relatively poor (Widen & Russell, 2008c).
Understanding individual differences in others’ emotion responding is also an important component of emotion understanding. Between the ages of 2 and 4 years, children show an increased ability to understand that other people may have desires and beliefs that differ from their own (Wellman, Cross, & Watson, 2001). Therefore they can begin to acknowledge individual differences in emotional responding based on personality, past experiences, and past emotions (Gnepp & Gould, 1985). Such understanding increases during middle childhood as children’s cognitive capacity to integrate information from various sources develops.
Most research focuses on when emotion understanding develops rather than how it develops. However, the principles of Eisenberg, Cumberland, & Spinrad (1998) socialization framework (i.e., modeling, contingent responding, and discussion) can also be applied to the development of emotion understanding. Regarding modeling, Holodynski and Friedlmeier (2006) have proposed that when infants imitate the emotion expressions they observe, they produce somatosensory and bodily experiences associated with the emotion and thus link these feelings with both the facial expression and the surrounding eliciting circumstances. Although there is little direct evidence that such a process occurs in infancy, the proposal is consistent with adult embodiment studies emphasizing the role of facial mimicry in emotion understanding (Niedenthal, Barsalou, Winkielman, Krauth-Gruber, & Ric, 2005). Regarding contingent responding, Holodynski and Friedlmeier (2006) also note that caregivers often imitate infants’ facial expressions. When caregivers do so, infants may develop associations between their own internal feeling states and the facial expressions shown by the caregiver (Holodynski & Friedlmeier, 2006). As indicated earlier, much research on emotion understanding has focused on discussion as a means of socialization. For example, Doan and Wang (2010) found that children’s emotion understanding is positively related to maternal discussion of other people’s thoughts and feelings in both European American and Chinese immigrant families.
Future Directions and Concluding Remarks
In this article, we attempted to provide a brief overview of contemporary theory and research on emotional development. Our overarching goal has been to inspire more efforts to bridge the gap between emotion scholars who focus on adults and those who focus on infants and children. We have argued that emotion theories based on research with adults must also be able to accommodate developmental data if they are to be deemed satisfactory accounts of human emotion. For example, appraisal theories of emotion elicitation merit testing in children and adolescents as well as in adults. At the same time, we believe that developmental researchers should attend more carefully to issues that are receiving considerable attention in the adult-oriented emotion literature. For example, much developmental research assumes the existence of discrete emotions as “natural kinds,” apparently unaware that the status of such entities is highly controversial within the adult literature (see L. F. Barrett, 2006). We hope that future developmental research will consider the implications of such controversies. In addition, we would suggest that greater awareness of the adult emotion literature might lead to new avenues of developmental investigation. For example, while considerable research has been conducted on facial expressions in infancy and adulthood, the natural history of their development during the interim periods of childhood and adolescence remains to be documented. In conclusion, we hope to encourage developmental and nondevelopmental scholars to reach out to each other in order to produce a new generation of research that can increase our understanding of both adult emotion and its developmental origins.