Child Recognition of Emotions
Disclaimer: This dissertation has been submitted by a student. This is not an example of the work written by our professional dissertation writers. You can view samples of our professional work here.
Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of UK Essays.
Human emotions serve as a means of expression, often indicating an individual's internal conscious experience or physiological arousal. Emotions also serve as a form of communication, alerting individuals to important aspects of their environment and their relationships with other individuals. Emotions influence a person's actions, cognitions, and how they are perceived by others (Strayer, 2002). For example, emotions may influence how individuals respond to an environmental threat, as feelings of sadness may indicate a withdrawal of behavior or feelings of fear may engage the activity of flight (Strayer, 2002). The experience of emotions, in turn, provides individuals with meaning about both their internal and external environments and information about how they should respond to a social situation.
The ability to develop an adequate understanding of emotions is known as emotional competence. More precisely, emotional competence is defined as "a demonstration of self-efficacy in emotion-eliciting social transactions" (Saarni, Campos, Camras & Witherington, 2006, p. 250). For children, one way to gauge their emotional competence is to examine their ability to perceive their own emotions and the emotions of others (Saarni et al., 2006). This type of assessment allows researchers to determine an approximate measure as to children's emotional development level.
The first skill of emotional competence is for children to develop an understanding of self. Through an ability to be aware of their own personal emotional experience, children begin to develop self-conscious emotions. Feelings of shame, guilt, and embarrassment, for example, allow children to refer to themselves as having conscious awareness that they are distinct from others (Lewis, 1993, 1995; Mascolo & Fischer, 1995). In addition, children's emotional competence is developed through an ability to be aware of multiple emotions or to feel that their emotions are in conflict with their environment (e.g., ambivalence; Stein, Trabasso, & Liwag, 2000). As children become aware of their own emotions, emotional development is strengthened and refined.
A second important skill development to emotional competence is for children to make sense of other's inner states (Saarni et al., 2006). Specifically, children learn to comprehend and interpret other's behaviors and begin to realize that others are capable of forming their own beliefs and emotions (for a review see Dunn, 2000; Halberstadt, Denham, and Dunsmore, 2001). Understanding the distinction between one's own emotions and the emotions of others is crucial to emotional development in children. For instance, studies indicate that children's ability to accurately identify emotions in self and in others may work as a gauge to assess social competence (Halberstadt et al., 2001). In these types of studies, children's social competence is established by correlating their understanding of emotion terms, facial expressions, and elicitors of emotion terms (e.g., situational descriptions of a happy or sad event) with their social competence rating from teachers' ratings or by peers' sociometric choices (Saarni et al., 2006). Children's ability to distinguish differences in emotional features in combination with how others view their ability provides a measure of their social competence. Monitoring children's social competence allows one to identify socially disadvantaged children and implement effective coping strategies before any harmful, long-term effects manifest (Benford, 1998).
Through children's awareness of their own emotional state, in combination with the skill to discern other's emotions, children begin to achieve more effective emotional processing skills. Understanding children's emotional processing is important because it affects many social outcomes, such as children's helping behavior (Chapman, Zahn-Waxler, Cooperman & Iannotti, 1987, Miller & Jansen op de Haar, 1997), aggressive responses (Harris & Siebel, 1975), and self-control (Ceschi & Scherer, 2003). Few studies, however, have examined how emotion affects children's abilities to accurately identify the emotional state of others.
Consequently, the present study sought to examine the effects of children's own emotional states on their social/cognitive abilities to recognize emotional states in others. More specifically, this research sought to understand how positive, negative, and neutral emotional states of children affected performance on emotion recognition tasks that utilized different levels of cognitive complexity. By utilizing two types of emotion recognition tasks, the research examined the influence of differently valenced emotions on children's social-cognitive abilities. Results may help to expand existing social information processing models by incorporating the influence cognitive complexity and affect may serve in children's recognition of others' emotions.
Understanding Other's Emotions through Facial Expressions
The ability for children to understand what others are experiencing emotionally develops through an interaction between the awareness of their own emotional experience and the ability to empathize and conceptualize the causes of emotions in others (Saarni et al., 2006). In addition, the more children learn about how and why others act the way they do, the more they can make inferences about the emotional state of others. Children typically rely on facial expressions to infer others' emotional state (Ceschi & Scherer, 2003; Holder & Kirkpatrick, 1991).
The face is considered the primary indicator of human emotion (Ekman, 1992). For example, body gestures are easily concealed (e.g., hiding a clenched fist behind one's back) or verbal communication can be eliminated by simply refusing to speak. Facial expressions, however, are more difficult to disguise (Ekman, 1993). Additionally, the diversity in an individual's face allows for a variety of emotional expressions, each associated with a distinct facial expression (Ekman, 1993). Facial expressions serve a dual purpose; facial emotions can indicate a person's internal emotional state or function as symbols referring to something else, such as a form of communication (e.g., deterring or placating someone's actions; Lewis & Michalson, 1985).
Facial expressions are commonly used as a means for gauging emotion states in research. By 2 ½ years, children can distinguish a number of basic emotional states in the facial expressions of adults, but do not always label them accurately (Izard, 1971). At 5 years, children can accurately label 41% of the emotions depicted in a set of adult photographs (Odom & Lemond, 1972). Overall, young children can recognize some of the more common emotional expressions as displayed by adults (e.g., MacDonald, Kirkpatrick & Sullivan, 1996).
By the ages of 11 or 12, most children recognize and verbalize that a person's expression may be both a social and an emotional response (e.g., Underwood & Hurley, 1999). Consequently, children realize that a person's facial expression may indicate both the individual's internal state (e.g., "I am feeling sad"), as well as what the cues represent socially (e.g., "I am expressing my feelings of sadness towards others"). Each form of emotional expression is essential in order for children to interpret and comprehend another's emotion (Underwood & Hurley, 1999).
Developmental Differences in Children's Understanding of Other's Emotional Experiences
As children mature, they acquire greater abilities to make inferences about what others are feeling (Gross & Ballif, 1991). Children, in an attempt to understand the emotions of others, begin to combine facial and situational cues. The ability to combine these cues, however, is strengthened and refined as children age. The easiest emotions for children to discern are positive ones (Saarni et al., 2006). Children can more readily identify happy reactions in a naturally occurring setting as compared to negative reactions (e.g. Fabes, Eisenberg, Nyman, & Michealieu, 1991). Negative facial expressions, on the other hand, such as sadness, fear, and anger, are more difficult for children to decode. Negative emotions become easier to interpret, however, when they are paired with an emotion-eliciting situational context (Saarni et al. 2006). In addition, the causes of negative emotion are easier for children to decode than causes of positive emotion, an explanation that appears consistent with negative emotions eliciting a more intense response (Fabes et al., 1991). For example, children can easily determine the causes for their goal failures because it is an undesired consequence.
Developmental differences are apparent when evaluating children's understanding of the causes of emotions (Fabes et al., 1991). Younger children (i.e., 3 year-olds) are more prone to attribute causes of emotion to a person's wants or needs, whereas older children (i.e., 5 year-olds) make use of other's personality traits to determine their future reactions to an emotional event (Fabes et al., 1991). Children aged 5 to 10 years can use a character's past experience to determine the character's reactions to a new situation (Gnepp & Gould, 1985). For example, if a character's best friend harasses him, children aged 5 to 10 can infer how that character will later react to seeing the best friend on the playground. The developmental difference is evident in the quality of the response. Younger children are more likely to infer what the character is feeling solely through the current situational information (e.g., the character would be pleased to see the best friend), whereas older children are more likely to use the prior experience to evaluate how the character will react (e.g., the character will avoid the best friend on the playground; Gnepp & Gould, 1985). Further support for this developmental difference is that younger children (i.e., preschoolers) are more likely to infer the emotional state of others when a character's emotional cues are presented explicitly (e.g., pictorial representation of the character's face) as compared to older children (i.e. school-aged) who can adeptly determine the character's response when less explicit cues are utilized (Lagattuta, Wellman, & Flavell, 1997).
These investigations demonstrate that by school age, children are well equipped to identify emotional expressions in others. Although there are developmental differences in children's abilities to identify reasons for the emotional expression, by the age of 5, children generally distinguish differences in emotional cues and identify different types of emotional expressions in others.
Integrating Cognition and Emotion
There has been considerable interest in how children interpret, encode, and respond to social environments. One such model that attempts to explain the relationship is the social information processing model (Crick & Dodge, 1994). The social information processing model assumes that the way in which children understand and interpret social situations directly influences how they respond behaviorally (Lemerise & Arsenio, 2000). In turn, the social information processing model offers an explanation for how children process and interpret cues in a social situation and arrive at a decision that facilitates their understanding of the social environment (Crick & Dodge, 1994; Dodge, 1986). For any social interaction, children utilize their past experiences and biologically determined capabilities (e.g., memory store capacity) in order to rapidly assess the situation (Crick & Dodge, 1994).
To illustrate the social information processing model, imagine a child who gets pushed on the playground by another child. First, the child must encode the social cues (both internal and external) to determine what happened (attention, encoding) and then determine why it happened (interpretation: an accident or on purpose?). In the third step of the model, the child begins to clarify his or her goal in the social situation (e.g., goal to show others he/she won't tolerate the behavior). In step four and five of the model, possible responses to the situation are generated in terms of anticipated outcomes and how those actions relate to the individual's goals (Lemerise & Arsenio, 2000). The child may choose to retaliate in response to the other child's actions or the child may choose to not retaliate for fear of the situation escalating. Finally, the majority of children generally choose the most positively evaluated response with respect to goals and anticipated outcomes before the behavior is enacted (e.g., the child ignores the push and walks away; Crick & Dodge, 1994).
The social information processing model has been useful in assessing how children encode and interpret social situations. The model, however, does not specify how emotion affects the processing strategy (Lemerise & Arsenio, 2000). Lemerise and Arsenio (2000) argue that it is possible to expand Crick and Dodge's model's explanatory power by integrating emotion processing with social information processing.
Before integrating emotion and social information processing, it is important to understand the relationship between the two. Emotions and cognitions may appear similar because both are types of information processing, but the way each influence human behavior makes them distinct (Lemerise & Arsenio, 2000). Emotion is about motivation; cognition, on the other hand, concerns knowledge. This view is shared by many functionalist theorists, neurophysiologists, and some cognitive theorists (e.g., Campos, Mumme, Kermoian, & Campos, 1994; Damasio, 1994; Oatley & Jenkins, 1996).
Because cognition and emotion are two distinct processes, an attempt to devise a model that integrates and utilizes the two is pragmatic. In Lemerise and Arsenio's (2000) social information processing model, the researchers added and expanded to Crick and Dodge's (1994) original concept. In particular, the researchers implemented other emotion processes that could influence accessing and evaluating responses. As an example of this approach, intense emotions can interfere with the steps of Crick and Dodge's model where children assess possible responses to a situation (Steps 4 and 5). For example, children with intense emotions may react negatively to a social situation (e.g., becoming easily upset and running away), thereby reducing the probability that they will interpret and encode the situation from the perspective of all parties (Lemerise & Arsenio, 2000). Intense emotions, in turn, can influence how a child responds in a social situation. In addition, the child's reaction to the social situation may be dependant on whether he/she cares about and wants that person to like him/her (Lemerise & Arsenio, 2000). Emotions heavily influence this social decision making process.
It is difficult to adhere to a social informational processing strategy without accounting for the influences emotion may serve. Support for the role emotions play in social information processing has been demonstrated in more recent research (e.g., Orobio de Castro, Merk, Koops, Veerman, & Bosch, 2005). Specifically, researchers examined the relationship between emotional aspects of social information processing and aggressive boys. After hearing a series of vignettes that instilled provocation by their peers, participants answered questions concerning social information processing, including feeling of their own emotions, the emotions of others, and emotion regulation. Aggressive boys used less adaptive emotion-regulation strategies, attributed more hostile intent to others' actions, and reported less guilt concerning their own actions (Orobio de Castro et al., 2005). For aggressive boys, anger attribution (i.e., encoding of emotions) significantly influenced the interpretation step of the social information processing model, a view that is consistent with Lemerise and Arsenio's (2000) model. Clearly, emotions can influence children's social information processing strategies. By combining emotional processing with social decision making processes, researchers can expand Crick & Dodge's model's explanatory power, perhaps offering further insight into the influence emotion serves for children's cognitive abilities (Lemerise & Arsenio, 2000).
Induction of Positive and Negative Affect
Researchers examine emotional influences on social information processing and other social behaviors by experimentally inducing emotions and assessing the effects (Bryan, Mathur & Sullivan, 1996; Bugental & Moore, 1979; Burkitt & Barnett, 2006; Carlson, Felleman & Masters, 1983; Masters, Barden & Ford, 1979; Stegge, Terwogt & Koops, 2001). Inducing affect typically consists of an experimenter having subjects recall events that make them happy or sad before examining their responses to a variety of social and cognitive problems. These problems can range from measures of altruism, self-gratification, or delay of gratification (Bugental & Moore, 1979). For this type of induction procedure, the researcher asks the child to recall and reflect upon a happy or sad past event for approximately 30 seconds to 2 minutes (Bryan et al., 1996). This type of procedure allows psychologists to examine how affective states influence individuals' social and psychological behaviors (Bugental & Moore, 1979).
Pre-recorded videotapes or audiocassettes are another technique used to implement positive or negative affect in children (e.g., Carlson & Masters, 1986; Rader & Hughes, 2005). The recordings improve the reliability and standardization of the affect inductions. The recording typically follows the same procedural method as the other affect induction studies (e.g., Carlson et al., 1983; Masters et al., 1979; Moore, Underwood, & Rosenhan, 1973)—the person reading the script (e.g., actor, puppet) asks the child to ruminate on a past experience that is positive, negative, or neutral for approximately 30 seconds.
For any procedural method chosen, it is important to validate if the affect induction actually takes place. There are multiple methods for conducting manipulation checks. Procedures include: having two or more experimenters rate the child's mood and assessing interrater agreement (e.g., Carlson & Maters, 1986); having participants use a word item check-list to indicate their current mood (e.g., Vosburg, 1998); or comparing if the performance of children in the positive or negative condition differs from those in the neutral condition (e.g., Bugental & Moore, 1979; Stegge et al., 2001). As demonstrated in past research, the induction of positive and negative moods is experimentally possible.
Positive and Negative Affect and Emotional Processing
There are a number of experimental studies that demonstrate the influence of children's emotional states on a variety of emotional processes and behaviors, such as altruism (Chapman et al., 1987, Miller & Jansen op de Haar, 1997), aggression (Harris & Siebel, 1975), and self-control (Ceschi & Scherer, 2003).
One study, in particular, induced positive emotional states in a group of 5-6-year-old children to examine their responses to social comparison situations where the participant was rewarded unfairly, sometimes in the participant's favor, sometimes in another's favor (Carlson & Masters, 1986). Children were exposed to one of three emotion inducing conditions: self-focused happy, other-focused happy, or neutral. After the children focused on their own happy emotional experience (self-focused) or the emotional state of a friend (other-focused) or had no emotional focus (neutral), they and other players received a reward for participating in a game. Children received either more (positive inequality) or less (negative inequality) of an award as compared to the other players. Children in the self-focused happy condition did not demonstrate a reduction in generosity after receiving an inequality of rewards (Carlson & Masters, 1986). The authors interpreted their results as supportive of the position that positive mood facilitates tolerance of aversive experiences (Carlson & Masters, 1986).
What these studies did not answer, however, is what influence emotion serves in other social information processes. Specifically, how do inductions of positive or negative affect influence children's emotion recognition?
One research experiment did attempt to investigate the influence children's own emotional states has on their ability to recognize emotions in others (Carlson et al., 1983). Experimenters induced emotions such as happiness, sadness, anger, or neutral affect in eighty 4 and 5 year-old children. The children were then provided with a label of an emotion (e.g. "happy") and asked to identify the correct facial expression from a group of photographs of other children who were displaying various emotions. Significant differences in accuracy across the mood induction conditions were not found, but children's own feelings of sadness did influence their perception of sadness in peers (Carlson et al., 1983). Sad participants were not more inaccurate than happy participants when identifying emotions, but when they were inaccurate they tended to confuse sadness with anger.
The induction of positive and negative mood in children appears to have an influence on children's emotional processing. Negative affective states appear to lead to inaccuracies in the perception of others' sadness, whereas positive affective states appear to help children maintain positive emotional experiences, even in the presence of aversive social situations.
Influence of Emotional States on Cognition for Adults
Despite the number of studies assessing children's emotion processing ability, only a few studies have investigated how emotion-directed information processes, such as perception, attention, judgment, and memory recognition and recall, are influenced by the child's own emotional state, whether enduring or temporary (Greene & Noice, 1988; Masters et al., 1979; Rader & Hughes, 2005). Because of this dearth in the literature, it is useful to review studies conducted with adults. One particularly important study, which examined the role affect plays in adult's cognitive performance, induced positive and negative affect through the use of a role-playing technique while participants carried out three cognitive tasks (Izard, Wehmer, Livsey, & Jennings, 1965). The cognitive tasks ranged from participants generating as many possible uses for a particular object (multiple-use task), recalling sets of numbers (digit span test), and giving verbalized responses for creativity problems. Positive affect increased performance for both the multiple-use task and the creativity problems as compared to those in the negative affect condition.
Some literature supports the finding that positive affect results in higher productivity and creativity (Ashby, Isen, & Turken, 1999). Specifically, induced positive affect improved cognitive processes such as memory, judgment, risk-preference, decision-making, creative problem solving, categorization, and logical problem solving (Ashby et al., 1999). Other research, however, does not support the claim that positive affect improves cognitive productivity and creativity. Positive affect may actually interfere with performance on some tasks (e.g., Kaufmann & Vosburg, 1997).
In an attempt to explain the discrepancy in these results, Forgas' (2000, 2002) affect infusion model (AIM) suggests that affect results in inattentive processing for complex tasks. As participants experience positive affective, for example, substantive processing or systematic processing may be hindered, thereby interfering with their ability to solve elaborate and complex problems. Specifically, negative moods may facilitate differentiated, analytic processing whereas positive moods may facilitate global, synthetic processing (Forgas, 2000). Negative moods may be more adaptive for cognitive tasks that require one to reduce complex decisions to a series of one-on-one comparisons, thus simplifying the results (i.e., analytic processing). Positive moods, on the other hand, may be more adaptive for cognitive tasks that require one to generate a wide variety of responses, often seeking out all possibilities for a solution (i.e., global processing).
Following Forgas' (2000, 2002) AIM model, it is clear how research supports the notion that positive and negative affect are adaptive for different types of cognitive tasks. To clarify, some studies show that positive affect facilitates cognitive performance by increasing participant's creativity (Ashby et al., 1999; Isen, Daubman & Nowicki, 1987; Isen, 2002; Izard et al., 1965). Other research, however, shows that positive affect results in inattentive processing, thereby reducing participant's creativity and problem-solving (Forgas, 2000; Kaufmann & Vosburg, 1997). These seemingly contrasting findings are explained by differences in task specificity. In the research conducted by Kaufmann and Vosburg (1997), for example, positive affect significantly inhibited creative problem solving. After the researchers experimentally induced affect, participants responded to a series of tasks presented in a "paper-and-pencil" format, and received no outside feedback. Contrastingly, in research conducted by Isen et al. (1987), they used creativity measures such as the "candle-problem," which required participants to physically manipulate objects and to come up with as many solutions to the problem as possible. In addition, the participants received feedback, which allowed them to instigate further solutions to the problem. Clearly, the tasks used in each of these studies are distinct. The notion that positive and negative affect are adaptive to different types of cognitive tasks is important because it points out the need to carefully consider the type of cognitive task being performed. Positive or negative moods may facilitate processing for different types of tasks in adults; therefore it is useful to examine how positive or negative moods affect children's processing in different types of tasks.
Influence of Emotional States on Cognition for Children
There are indications that the influences of positive affect on children's cognitive performance are similar to those in adults (Rader & Hughes, 2005). For example, eighth-grade students who were experimentally induced with positive affect showed greater cognitive flexibility than students in the control condition and obtained higher scores on a verbal fluency test (Greene & Noice, 1988). Likewise, researchers have examined the effects of emotional states on learning (Masters et al., 1979). After the induction of a positive, negative, or neutral emotional state, children completed a series of shape discrimination tasks. The dependent variable in the experiment was how many trials it took the children to achieve perfect mastery for the task (e.g. identifying 12 consecutive trials of shapes correctly). For children in the positive affect condition, positive affect enhanced performance. Contrastingly, for children in the negative condition, negative affect hindered performance dramatically (Masters et al., 1979). In addition, positive affect increased performance for children on a block design task, a challenging cognitive task that requires the use of spatial analysis (Rader & Hughes, 2005).
Research also suggests a relationship between affect and children's thinking processes (Bryan et al., 1996). Specifically, negative affective states decrease participant's efforts for processing cognitive information (Ellis, Thomas, & Rodriquez, 1984). Positive affective states, on the other hand, improve participant's memory on various tasks, which include: mastery of a discriminatory task (Masters et al., 1979); altruism (Chapman et al., 1987; Miller & Jansen op de Haar, 1997); and child compliance (Lay, Waters & Park, 1989). In sum, positive affective states increase complex cognitive functions when participants are required to synthesize information in new and useful ways (e.g., word association and memory tasks, creativity tasks, problem-solving tasks; Bryan et al., 1996).
Social Information Processing and Cognitive Complexity
Children's awareness of their own emotional state, in combination with their skill to discern other's emotions, allows them to develop more effective social information processing skills. As children become more aware of emotions they or others are experiencing, it facilitates problem-solving (Saarni et al., 2006). In turn, when children know how to respond emotionally to an encounter, it can aid in their decision making strategy, thus influencing behavioral or cognitive processing strategies.
Task complexity can negatively influence accuracy in identifying emotional expressions in others (MacDonald et al., 1996). Specifically, research has shown how incorporating contextual information for an emotion recognition task results in lower levels of performance as compared to a task where children are given the label for the emotional expression (MacDonald et al., 1996). Labeling tasks involve an extremely easy stimulus (i.e. children are given a word), whereas contextual information tasks involve integrating and synthesizing implicit information (i.e. children must derive a word from the vignette)plexity. paragraph should be eliminated. This is more relevant to cognitive processing strategies rather than levels of task. Adding contextual information to an emotion recognition task, therefore, increases the difficulty of the task, resulting in lower performance, especially for younger children (MacDonald et al. 1996).
Past research on children's emotional recognition has not adequately addressed the influence of children's own emotional states (positive or negative) on the accuracy of the perception of emotional states in others. Research addressing the topic is minimal; only a few studies have approached the issue (e.g., Carlson et al., 1983). In addition, past research has not directly demonstrated how emotion and cognitive task complexity influence children's ability to recognize emotion in others.
Based on the information regarding children's ability to recognize emotions in others through their facial expressions; the developmental differences in children's facial recognition abilities; the influence of emotion on children's emotion processing; and the influence of emotion on adult's cognitive processing, researchers can devise an appropriate social information processing model. The model, in effect, should integrate emotion and cognitive processes to determine the influence affect and task complexity have on children's recognition of emotions in others.
The component of the model the present study investigated is how children's own emotion affects their interpretation of social cues, specifically the emotional expression of others. Inaccurate interpretations will provide potential consequences to children's subsequent social decision making processes.
The Present Study
The study examined the influence positive and negative affect has on children's emotion recognition. Children, aged 5-to 8-years, participated because of their ability to identify emotions in others (Fabes et al., 1991, Gnepp & Gould, 1985, Saarni et al., 2006). Because emotion processing and cognition are considered an integral part of children's social competence (Lemerise & Arsenio, 2000), the study design combined and evaluated cognition and emotional processes. Specifically, the study investigated how the cognitive complexity of the task interacted with mood effects on emotion recognition performance.
In the experiment, children were individually tested. They were first exposed to one of three mood induction conditions (positive, negative, or neutral) using a computer setup with a pre-recorded audio file, a method that is consistent with a brief mood induction procedure (Rader & Hughes, 2005). For the testing procedure, the experimenter utilized two forms of emotion expressing questions: label-based vignettes and context-based vignettes. For the label-based vignettes, an audio file of an experimenter naming an emotion (e.g., happy) was played. For the context-based vignettes, an audio file of an experimenter reading a vignette, which subtly conveyed an emotion, was played. Following both the label-based and context-based vignettes, the experimenter then asked the children to indicate which face best fit the emotion indicated. A male or female face with his/her four different emotional expressions (happy, sad, surprise, disgusted) was displayed on the computer monitor. Children responded to each question by clicking with a computer mouse on the face they believed best fit the emotion. The format of the questions was intended to help answer how children's affect influences performance on tasks of different cognitive complexity.
Previous research indicates that the label-based task, a task in which children are given a specific emotion, produces a higher rate of accuracy in identifying emotion in faces as compared to the context-based task, a task in which children derive the emotion from contextual information (MacDonald et al., 1996). Previous research also indicates that positive affect increases complex cognitive functions when participants are required to synthesize information in new and useful ways, characteristics necessary to correctly identify contextually-based questions (Bryan et al., 1996). Contrastingly, negative affect has been shown to produce lower-level cognitive processing efforts, which may be detrimental to performance on more difficult cognitive tasks (Leight & Ellis, 1981). Negative affect, in addition, has been shown to promote inaccuracy for emotion recognition tasks and produce longer response times (Carlson et al., 1983). Lastly, children in a sad affect condition tend to confuse sad emotional expressions for angry emotional expressions when they are inaccurate, indicating that children may systematically misidentify negative emotional expressions when negative affect is induced (Carlson et al., 1983).
Based on previous research, the present study predicted that 1) children would perform at a significantly higher rate of accuracy and have quicker response times for emotion recognition in the label-based questions as compared to the context-based questions, regardless of affect condition; 2) for children in the positive affect condition, accuracy would be significantly higher and response times would be significantly quicker for context-based questions as compared to children in the negative affect condition; and 3) for children in the negative affect condition, they would systematically misidentify negative emotion labels within the category of negative valence (e.g., confusing sadness for disgusted). Differences in performance based on children's ages were also expected. It was predicted that older children (i.e., 7-8-year-olds) would respond more quickly and accurately than younger children (i.e., 5-6-year-olds). Children in the neutral condition served as a control.
Data included responses from 87 children aged 5 (n = 37), 6 (n = 23), 7 (n = 16) and 8 (n = 11) years of age (45 females, 42 males). The number of participants was determined by conducting a power analysis using the program G*Power with a moderate effect size of .35, alpha of .05, and power of .85 (Erdfelder, Faul,, & Buchner, 1996). The race/ethnicity of the children was fairly representative of the local population: 2.3% Africa-American; 64.4% Caucasian; 10.3% Latin-American; 19.6% Multi-racial or other; 2.3% Pacific-Islander; and 1.1% did not report race. Data were collected for an additional 10 children, but were excluded for the following a priori reasons: off-task (n = 3); developmental delays (n = 1); chose to stop the study (n = 4); and experimenter error (n = 2). To adjust for equal sample sizes across affect conditions, an additional seven participants' data from the neutral mood induction condition were chosen at random to be excluded from the data analysis. Data from 87 children (29 = positive condition, 29 = neutral condition, 29 = negative condition) comprised the final analyses.1
Research assistants determined participant's eligibility using a database of birth announcement records that the Baby and Child Rebel Lab maintain. Research assistants also actively recruited participants through a local pre-school center on the UNLV campus.2 Letters describing the study were mailed to the parents of the children and follow-up phone calls were made to schedule appointments. Children received a small gift for participating in the research after the experiment.
Stimulus faces. The stimuli consisted of 48 digitized, color images of 12 adults (6 female, 6 male), with four images of each adult displaying a different facial expression: happy, sad, surprised, and disgusted. The adult images included four African-American/Black, four Caucasian, and four Hispanic/Latino/Spanish adult faces, 10 of which were taken from the NimStim Face Stimulus Set (Tottenham, Borscheid, Ellertsen, Marcus & Nelson, 2002).
The NimStim Face Stimulus Set consists of 657 photographs of actors of different ethnicities posing neutral, angry, calm, disgusted, fearful, happy, sad, and surprised facial expressions (Tottenham et al., 2002). These photographs were rated by US-born college students. Using Cohen's kappa, agreement for the facial expression ratings maintained at least a 70% criterion, meaning that at least 70% of the observers judged the emotion to be the one determined by the researcher. The NimStim Face Stimulus Set (Tottenham et al., 2002) has been used and validated in previous research (Kirsh & Mounts, 2007; Palermo & Coltheart, 2004).
For the remaining images used in the study, the faces were obtained from a photo database maintained by our lab. Forty adults (25 females, 15 males) rated the faces for four facial expressions (happy, sad, surprised and disgusted faces) on a 5-point scale (1 = no sign of the given emotion, 2 = vague, ambiguous sign of the emotion, 3 = clear display of the emotion but not a very strong display, 4 = relatively stronger display of the emotion, and 5 = exceptionally strong, sustained display) (Eisenberg, Fabes, Bustamante, Mathy, Shell & Rhodes, 1988; Eisenberg, Schaller, Fabes, Bustamante, Mathy, Shell & Rhodes,1998). There was high interrater agreement for the faces across the happy (alpha = .96), sad (alpha = .94), surprised (alpha = .99), and disgusted (alpha = .97) emotional expressions.3 The average ratings for the facial expressions were: happy (M = 4.36, SD = .76); sad (M = 3.21, SD = 1.01); surprised (M = 3.20, SD = .96); and disgusted (M = 3.30, SD = 1.18). Researcher assistants used Adobe Photoshop, to standardize all facial images to be of similar size, brightness/contrast, and background.
Mood induction. To ensure consistency across subjects, a researcher constructed three audio files of the experimenter reading the elaborative mood induction scenarios, one positive, one negative, and one neutral. During the study, the audio files were played through computer speakers to the participants. All three scenarios have been used in previous research with similar-aged children. The negative mood induction script was originally developed to test 6-year-olds' self-control with emotional reactions (Meerum-Terwogt, Schene, & Harris, 1986). A similar positive mood induction script, which is balanced with an equal number of emotional descriptors as the negative mood induction script, was later developed and validated to examine the effects of affect on the size of children's drawings (Burkitt & Barnett, 2006). To alleviate children's mood in the negative affect condition, children heard a mood-repair script following the experiment (Meerum-Terwogt et al., 1986). The neutral mood induction script is a list of table manners for children, similar to a script used in previous research (e.g., Rader & Hughes, 2005). See Appendix 1 for the negative, positive and neutral mood induction scripts and the mood-repair script.
To examine the effectiveness of the elaborative mood induction procedure, researchers used The Facial Affect Scale (Careseo & Cohen, 1998) after testing was completed, but before presenting the mood-repair script in the negative mood induction condition. The materials are a set of five schematic faces that express emotions ranging from very happy to very sad. Children circled which face best represented how the story made them feel. The Facial Affect Scale has been used and validated in previous research with similar-aged children (e.g., Rader & Hughes, 2005). Likewise, the length of the mood induction procedure utilized in this study, approximately four minutes, is sufficient in sustaining the children's mood for the length of the testing procedure (e.g., Bugental & Moore, 1979; Burkitt & Barnett, 2006, Carlson et al., 1983; Masters et al., 1979; Rader & Hughes, 2005; Stegge et al., 2001).
Emotion recognition testing. For the testing procedure, the experimenter utilized two forms of emotion expressing questions: label-based and context-based. The context-based vignettes covered four emotional categories: happy, surprised, sad, and disgusted. Each vignette was designed to examine emotion recognition in children (Ribordy, Camras, Stefani, & Spaccarelli, 1988). See Appendix 2 for the vignettes.4 The context-based questions served as a type of task that required making an inference about the emotion based on past experiences/social interactions (Ribordy et al., 1988). Participant's responses to the happy, sad, surprised and disgusted questions were utilized in the data analysis. These vignettes were validated in research investigating 5-and 6-year-olds' ability in choosing the emotion that best matched the emotion of the character in the vignette (MacDonald et al., 1996).
The label-based questions involved the experimenter naming an emotion and participants identifying the face that best matched the descriptor. These questions were used successfully with 6-year olds in other research (e.g., Carlson et al., 1983; MacDonald et al., 1996). The emotions for the label-based questions also included happy, sad, surprised and disgusted. The label-based questions served as a type of task that required matching a label to an expression (a matching task), a type of task that is less cognitively complex for children, producing a higher rate of accuracy as compared to context-based questions (MacDonald et al., 1996).
Setup and equipment. Participants were seated approximately 3 feet away from a 15" computer monitor that displayed the stimulus faces. The researchers presented stimuli on the screen of a Macintosh iBook computer while the audio stimuli played through computer speakers. The experimenter used SuperLab to display each set of trials and record responses and reaction times. The experimenter sat in the room with the child for the duration of the study in order to explain the directions to the child and to monitor responses.
When the participant arrived, a trained experimenter obtained consent from the parent and written assent from any child aged 7 years or older. While parents completed the demographic form, the experimenter introduced the child to the Facial Affect Scale. Similar to previous research procedures (Rader & Hughes, 2005) children colored in the five faces as the experimenter pointed to each face in a randomized order and provided a specific label for which emotion it expressed (e.g. very happy, happy, calm, sad, very sad).
Following the completed paperwork, children were brought into a separate room to complete the study. Parents could observe their child via a television screen while in the waiting area. When the child first entered the testing room, the experimenter gave instructions to the child on how to use the computer and mouse. The purpose in introducing children to the computer and mouse was to ensure children were comfortable using the equipment before the study began (children had little to no trouble using the computer mouse). The experimenter explained how to complete the study to the children through the use of six practice trials (three context-based practice trials, three label-based practice trials) as the child practiced clicking on one of four schematic faces displayed on the computer monitor for each trial. The explanation and practice trials lasted for approximately three minutes. Following the practice trials, the child was randomly assigned to one of three mood-induction conditions (positive, negative, or neutral) and listened to the script via the computer speakers (See Appendix 1 for mood induction scripts). Children from the local preschool as noted2, did not receive random assignment to conditions; children only participated in the positive mood induction condition (n = 10). The scripts lasted approximately four minutes.
Following the mood induction procedure, the experimenter asked the children the label-based and context-based questions. The order of presentation for these two formats was counterbalanced with approximately half of the participants receiving the label-based questions first and approximately half of the participants receiving the context-based questions first.
For the label-based question procedure, the experimenter read a brief set of instructions to the participant before any of the faces were shown (adapted from Ribordy et al., 1988). The instructions detailed what the participant would see on the computer screen and how they would respond to the questions. See Appendix 3 for instructions. Next, the experimenter presented six trials of faces on the computer screen. For each trial, children first heard a specific emotion (e.g., happy, sad, surprised, or disgusted) labeled and were asked to identify the specific emotion (i.e. "Sad. Which face is the sad face?"). Immediately following the audio file, four faces of the same individual with different emotional expressions (happy, sad, surprise, and disgust) were displayed on the screen. Two of the photographs were presented horizontally at the top of the screen and two horizontally at the bottom of the screen. The ordering of the faces was randomized for race/ethnicity and gender for each of the trials.
For the context-based questions, the experimenter read a brief set of instructions to the participant before any of the faces were shown. See Appendix 3 for instructions. Next, the experimenter played the first audio file, a short vignette that subtly conveyed an expressed emotion followed by a question asking the participant to choose the face that best fit the vignette ("It's Christmas morning and Jane/John got all the presents he/she wanted and he/she hugged his/her family. Which face best fits my story?"). Following the audio file, four faces of the same individual with different emotional expressions (happy, sad, surprise, and disgust) were displayed on the screen.
For both the label- and context-based questions, children were asked to identify one emotion per trial. Once the child clicked the mouse on one of the four faces, the study proceeded to the next trial. If the participant did not respond after 1-minute, the experimenter continued to the next trial and responses for that trial were deleted (n = 14 trials). There were an equal number of positive (happy, surprised) or negative (sad, disgusted) emotion questions throughout the label-based and context-based question trials. Children heard and responded to a total of 12 trials (6 label-based and 6 context-based), a length of time that is adequate to maintain children's attention spans (Carlson et al., 1983; MacDonald et al., 1996).
Following the presentation of the label-based and context-based questions, children completed the Facial Affect Scale. Children were asked to indicate how the story they listened to at the beginning of the study made them feel. Children responded by coloring in one of the five schematic faces that ranged from very sad to very happy. For children in the negative condition, following their response to the Facial Affect Scale, the experimenter then played an additional audio file that served as a mood-repair. See Appendix 1 for the mood repair script. The entire procedure lasted approximately 30 minutes.
Mood Manipulation Check
We used a one-way ANOVA to examine the effectiveness of the mood induction condition (positive, neutral, and negative) on children's self-reports for how the story made them feel, as indicated by the Facial Affect Scale (1 = very sad, 5 = very happy). Our analysis indicated that there was a successful manipulation of mood between groups, F(1, 2) = 26.87, p < .001. Children in the positive condition (M = 4.28, SD = 1.16) and the neutral condition (M = 4.48, SD = .785) reported significantly higher positive emotions than those reported by children in the negative condition (M = 2.38, SD = 1.54).
Data Analysis Procedure
An analysis was conducted to determine if the order in which children received either of the question types (e.g., label-based or context-based questions received first) influenced children overall percent of correct responses or response times. The analysis indicated that order was not a significant influence on children's percent of correct responses or response times. Data, therefore, were collapsed across the variable, order, in all subsequent analyses.
We used a SAS PROC MIXED analysis with repeated measures to determine 1) children's accuracy based on question type (label-based vs. context-based), mood induction condition (positive, neutral, negative) (mood induction condition will be henceforth referred to as mood), and emotional expression (happy, sad, surprise, disgusted); and 2) children's reaction times based on question type, mood and emotional expression.5 A SAS PROC MIXED analysis assumes that the data are normally distributed, however, the analysis does not require the data to be independent or maintain constant variance across groups (i.e. offers more flexibility than ANOVA). In addition, a SAS PROC MIXED analysis is able to include subjects with multiple responses that have one or more data values missing at random (Littell, Milliken, Stroup & Wolfinger, 1996). Post hoc analyses were conducted using differences in least-squares means with Tukey-Kramer's adjustment for all analyses. Children's age was included as a between-participants variable for all analyses.6
A logistic regression analysis was conducted to determine 3) if age, mood, question type, or emotional expression predicted inaccuracies in the identification of other's emotional expressions. Inaccurate interpretations may provide potential consequences to children's social decision making processes. Lastly, a chi-square analysis was also conducted to determine 4) if children were more likely to make errors within a same category emotion (e.g. confusing happy for surprise) than errors in a different category emotion (e.g., confusing happy for sad).
Percent of Correct Responses
The dependent variable used for the analysis was percentage of total questions answered correctly. There was a main effect for question type, F(1,75) = 41.31, p < .0001. Accuracy was significantly greater for label-based questions (M = .91, SD = .28) than for context-based questions (M = .72, SD = .45). In addition, there was a significant main effect for age, F(3,75) = 9.67, p < .0001. The older the children, the more accurate they were (See Figure 1).
Post hoc analyses revealed significant differences between 5- and 7-year-olds, t(75) = -3.85, p < .001, 5- and 8-year-olds t(75) = -4.61, p < .001, 6- and 7-year olds t(75) = -2.02, p < .05, and 6- and 8-year-olds, t(75) = -2.97, p < .01.
There was a main effect for emotional expression, F(3, 225) = 3.98, p < .01. Of the four emotions used in this experiment, accuracy was greatest for sadness, followed by disgust, happiness, and surprise (in descending order), ranging from 86.3% accuracy for sadness to about 77% for surprise (See Figure 2).
Post hoc analysis revealed significant differences between the emotions of happy and sad, t(225) = -3.11, p < .01, happy and disgusted, t(225) = -2.48, p < .05, and sad and surprised, t(225) = 2.10, p < .05.
The main effect of emotional expression was superseded by a significant two-way interaction between question type and emotional expression, F(3, 199) = 3.28, p < .05 (See Figure 3).
Post hoc analysis revealed significant differences for percent of correct responses between label-based and context-based questions for the emotional expression of happy, t(199) = 5.76, p < .01, sad, t(199) = 2.09, p < .05, and surprised, t(199) = 3.32, p < .05. In addition, for the context-based questions, significant differences were found between percent of correct responses for happy and sad, t(199) = -3.96, p < .01, happy and disgusted, t(199) = -3.67, p < .01, and sad and surprised emotional expressions, t(199) = 2.25, p < .05.
The dependent variable used in the analysis was the average length of time (in ms) that it took for children to respond to the questions. Similar to previous research with reaction time data on children's responses to emotional expressions (Hadwin, Donnelly, French, Richards, Watts & Daley, 2003), extreme scores (scores falling 2.5 standard deviations away from the mean) were excluded from the analysis (n = 3).
There was a main effect for question type, F(1,73) = 19.90, p < .0001. Response times were significantly quicker for label-based questions (M = 4652.8, SD = 3109.1) than for context-based questions (M = 6629.8, SD = 5925.7). A main effect for mood induction condition was also found, F(2,73) = 4.31, p < .05 (See Figure 4).
Post hoc analyses indicated children in the positive mood condition were significantly quicker than children in the negative mood condition, t(73) = -2.93, p < .01. Differences in response times between positive and neutral mood conditions approached significance, t(73) = -1.75, p > .08.
There was also a main effect for age, F(3,73) = 12.55, p < .0001. The older the children, the quicker they completed the tasks (See Figure 5).
Post hoc analyses revealed that the 5-year-olds responded more slowly than all the other age groups, ps < .05. The 6-year-olds responded more slowly than the 8-year-olds, t(73) = 2.53, p < .05. The 6- and 7-year-olds and the 7- and 8-year-olds did not differ in response times.
Systematic Inaccuracies in the Identification of Others' Emotional Expressions
If children misidentified emotions, but were able to categorize the expressions as positive or negative, they should be more likely to make errors within the same category than errors in different categories (e.g., confusing sadness for anger as seen in Carlson et al., 1983). To examine whether age, mood condition, emotional expression, and question type affected children's inaccuracies in their identification of emotions, a step-wise logistic regression analysis was conducted. Children's mislabeling of same category emotions (i.e., confusing happy for surprise or confusing sad for disgusted) or different category emotions (i.e., confusing happy for sad) was used as the dependent variable for the analysis. There were no significant results.
The logistic regression analysis revealed that none of the factors used in the analysis affected children's inaccuracy of response. As a result, the data were collapsed across all variables in order to examine the likelihood of children making errors within same and different category emotions. A chi-square analysis, therefore, was conducted to examine if children had a higher frequency of error for within the same category emotions (e.g. confusing happy for surprise) as compared to different category emotion (e.g., confusing sad for happy). Results indicated that children were more likely to make errors within the same category emotion (60.4%) than they were to make errors for the different category emotions (39.6%), ?2(1, N = 192) = 8.33, p < .01.
Results demonstrated that children were more accurate and had faster response times on the label-based questions as compared to the context-based questions, regardless of their affective state. In addition, the older children were, the better their performance on both types of cognitive tasks (label-based and context-based questions). Younger children were less accurate and took longer to respond to the questions than the older children.
Mood appeared to have a limited influence on children's responses. Positive mood facilitated shorter response times, but did not significantly influence children's accuracy. In addition, question type influenced accuracy for the identification of emotional expression; for the context-based questions, children were more accurate at identifying negative emotional expressions (e.g., disgusted) as compared to positive emotional expressions (e.g., happy). Children were also more likely to make same category emotional expression errors (e.g., confusing happy for surprise) than different category emotional expression errors (e.g., confusing happy for disgust).
Cognitive Difficulty of Question Type
Results supported the first hypothesis, which predicted that children would perform at a higher level of accuracy and have quicker response times for the label-based
questions as compared to the context-based questions. A likely explanation for the results is the difference in complexity of the context vs. label questions. Contextual information can aid children's understanding of complex emotions in an identification task, specifically for emotions like disgust (Ribordy et al., 1988). Sometimes, however, when contextual information is added to an emotion recognition task, it decreases children's accuracy of response (MacDonald et al., 1996). Contextual information may increase the cognitive complexity of the task, thus resulting in lower performance as compared to label-based tasks. The use of contextual information therefore may assist children in determining how someone might feel, thus serving a facilitative purpose (e.g. Carroll & Russell, 1996). However, inferring an emotion from context is much more complex than simply attaching an emotional expression to a label. One would expect a lower level of accuracy, therefore, for context-based questions compared to label-based questions (MacDonald et al., 1996). In addition, one would expect slower response times for tasks that are considered more complex and challenging (Masters et al., 1979)
Labeling tasks involve a simple stimulus (i.e. children are given a word) whereas contextual information tasks involve integrating and synthesizing implicit information (i.e. children must derive an emotion from the vignette)plexity. paragraph should be eliminated. This is more relevant to cognitive processing strategies rather than levels of task. More specifically, context-based questions require children to make an inference about an individual's emotion. Inferring emotional expressions, therefore, requires knowledge about emotional expression labels (Bennett, Bendersky & Lewis, 2005), what emotional expressions look like (Ekman, 1992; Ekman, 1993), and factors that might influence emotions (Carroll & Russell, 1996; Gross & Ballif, 1991). Labeling tasks, on the other hand, do not require an individual to consider factors that might influence a person's emotion. Moreover, labeling tasks do not require children to make an inference; rather it requires children to simply match labels and expressions. Inferences rely on recalling memories about emotion because the emotional expression is not explicitly presented to the child. Matching labels and expressions simply requires recognition, a type of task that requires less cognitive capacity (Whiting & Smith, 1997). As found in the present study, adding contextual information to an emotion recognition task increased the difficulty of the task, resulting in lower performance (MacDonald et al. 1996).
Positive Mood Induction
Results did not fully support the second hypothesis, which predicted that children in the positive mood induction condition would perform at a higher rate of accuracy and with quicker response times on the context-based questions as compared to children in the negative mood induction. Specifically, mood did not affect accuracy overall or for either question type. The induction of mood, however, had a substantial influence on children's rate of response for the questions. Children in the positive mood induction condition completed the label-based and context-based questions significantly faster than children in the negative mood induction conditions.
Accuracy of emotion identification appears to be heavily influenced by children's emotion knowledge, which is often guided by experience (Bennett et al., 2005). Correctly identifying the emotional expression of others, as a result, would be dependent on children's maturity; increases in social experiences can lead to increases in emotion knowledge. Results from the present study, therefore, are aligned with previous research, indicating that the induction of affect cannot substitute for emotion knowledge experience. The induction of affect, as a result, will not increase children's accuracy of response for emotion identification. Children's affective social competence, therefore, remains as the best predictor for how well they can correctly identify emotional expressions, especially as their affective social competence develops and becomes more complex through maturation and increases in experience (Halberstadt et al., 2001).
Contrastingly, the rate at which children identify an emotional facial expression is influenced by the induction of positive affect. Affect can influence children's response rates in the identification of emotions in others (Carlson et al., 1983, Masters et al., 1979). Positive affect, for example, increased children's rate of responding, providing shorter response times (Masters et al., 1979). The induction of negative affect, in contrast, can impair children's response rates, causing longer reaction times (Carlson et al., 1983). Results from the present study, therefore, are consistent with previous research.
Results from the present study showed that positive affect increased children's rate of cognitive processing, helping them to reach a decision concerning the emotional expression of others more quickly (Baumann & Kuhl, 2005; Bryan et al., 1996). Ease of cognitive processing, in turn, may directly benefit from the induction of positive affect. Previous research has indicated a potential link between cognitive processing and positive affect. Easy-to-process stimuli (e.g. pictures of line drawings) are more likely to produce expressions of positive affect in adult participants (Winkielman & Cacioppo, 2001). Recognizing a familiar face (an easy to process stimuli), therefore, will most likely elicit a positive evaluative response, which can be measured through self-reports and psychophysiological measures (e.g. participant will smile). As a result, a connection between fluency of processing and feelings of positive affect is suggested (Winkielman, Schwarz & Nowark, 2002).
Consistent with previous research, results from the present study demonstrated a connection between ease of processing and positive affect. Results, however, suggested that the link between ease of processing and affect may be bi-directional. Ease of processing may induce positive affect (Winkielman & Cacioppo, 2001), but the induction of positive affect may also facilitate cognitive processing.
The ability for positive affect to decrease participants' response times for various cognitive tasks has been clearly demonstrated in research conduc
Cite This Dissertation
To export a reference to this article please select a referencing stye below: