Faculty of Social and Behavioural Sciences
Graduate School of Child Development and Education
Autistic Traits and Social Emotions in Early
Childhood: The role of Socio-Cognitive
Ability
Research Master Child Development and Education Research Master Thesis
Shanna van Trigt (10365664)
Supervision: Dr. Milica Nikolić, Dr. Cristina Colonnesi, Dr. Bonny van Steensel, & Prof. dr. Bram Orobio de Castro
Abstract
Children with autistic traits may experience dysregulated social emotions, which could lead to difficulties in social functioning. Impaired socio-cognitive abilities due to higher levels of autistic traits, may play a role in the dysregulated social emotions. We examined associations between autistic traits and social emotions, and the role of socio-cognitive abilities. We observed 2- to 5-year-old children (N = 98, 50% girls, Mage = 4.06) during the Broken Toy
task in which they were led to believe they broke the experimenter’s favorite toy to elicit guilt, embarrassment, and shame-like avoidance. Parents reported about children’s autistic traits, using the Social Responsiveness Scale. Socio-cognitive abilities were measured with the Theory of Mind-test-Revised and emotion recognition task. Children with higher levels of autistic traits had lower levels of socio-cognitive abilities, which, in turn, were related to less guilt and embarrassment expressions after a trangression. Higher levels of autistic traits were directly related to less embarrassment, and more guilt and verbal shame-like avoidance. Thus, dysregulated social emotions are related to autistic traits already in early childhood and in a non-clinical sample. Since well-regulated social emotions are important for social
relationships, this study offers important information for prevention programs targeting children at risk for autism.
Autistic Traits and Social Emotions in Early Childhood: The role of Socio-Cognitive Ability Social emotions, or self-conscious emotions, are emotions that arise in relation to social rules, and are self-evaluative in context of these social rules (Davidson, Vanegas, & Hilvert, 2017). It has been assumed that these emotions require an understanding of oneself as a social object (self-awareness), of the self in relation to social rules, and of others’
evaluations of one’s behavior (Lewis, 1993, 1997). Children usually develop self-awareness around the age of 2 years and incorporate social rules and norms necessary to evaluate their own behavior around the age of 3 years (Lewis, 1997, 2007). Some social emotions, such as guilt and embarrassment, are considered to be adaptive, as they motivate prosocial behaviors and evoke forgiveness and appeasement in others (Tangney, 1999; Tangney, Miller, Flicker, & Barlow, 1996). Other social emotions, such as shame, are considered to be less adaptive, as they inhibit prosocial behavior by withdrawal and avoidance (Feinberg, Willer, & Keltner, 2012; Tangney, 1999; Tangney et al., 1996). Dysregulations in social emotions could lead to problems in forming and maintaining social relationships, since social emotions may drive individuals to show either adaptive or maladaptive behaviors (Muris & Meesters, 2014; Muris et al., 2016; Tangney, 1999).
The dysregulation of social emotions may play a role in social functioning in children with autism spectrum disorder (ASD). ASD is a neurodevelopmental disorder associated with severe and pervasive deficits in social interaction and communication (Schaller & Rauh, 2017). Children with ASD tend to have difficulties in forming and maintaining social
relationships, which may, at least in part, be due to abnormalities in social emotions (Hobson et al., 2006). Several studies indicated associations between dysregulations in the expression of social emotions and ASD (Davidson, Hilvert, Misiunaite, & Giordano, 2018; Davidson et al., 2017; Harms, Martin, & Wallace, 2010; Hobson et al., 2006; Muris et al., 2016; Schaller & Rauh, 2017). These dysregulated social emotions in children with autism may be explained
by deficits in socio-cognitive abilities, such as Theory of Mind (ToM: Davidson et al., 2018, 2017; Hobson et al., 2006; Schaller & Rauh, 2017).
Having high levels of autistic traits in early childhood may be the first sign of developing ASD later in life (Cox et al., 1999; Volkmar, Chawarska, & Klin, 2005). In addition, as mentioned earlier, dysregulated expressions of social emotions poses risks for developing problematic social relationships (Muris & Meesters, 2014; Muris et al., 2016; Tangney, 1999). However, so far, little is known about social emotions in early childhood (Drummond, Hammond, Satlof-Bedrick, Waugh, & Brownell, 2017). Additionally, research that addresses the relation between social emotions and autistic traits in early childhood is lacking. Pre-school children with high levels of autistic traits showed an absence of coy smiles—which are indicative of shyness or embarrassment (Draghi-Lorenz, Reddy, & Costall, 2001; Reddy, 2000), when looking in the mirror, unlike pre-school children with Down syndrome and typically developing children, who both did show coy smiles (Reddy, Williams, Costantini, & Lang, 2010). However, no studies that address autistic traits in relation to guilt and shame in early childhood have been found. Nevertheless, several studies stress the importance of early treatment for improving social behavior in children with high levels of autistic traits (Corsello, 2005; Cox et al., 1999; Rogers, 1996; Rogers & Lewis, 1989). Thus, knowing if, and why there are dysregulations in social emotions in young children with high levels of autistic traits may be of importance. The present study examined the association between autistic traits and social emotions, and the role of socio-cognitive abilities, in early childhood.
Social Emotions
The first social emotion that this study addresses is guilt. Guilt involves a sense that one’s own displayed behavior—a specific action—was bad, but it does not inflict a negative evaluation of the whole self (Barrett, 2005; Drummond et al., 2017; Lewis, 2007; Muris,
2015; Muris & Meesters, 2014; Tangney & Dearing, 2002). If experiencing guilt, one regrets the action or behavior and wished they behaved differently. Therefore, feelings of guilt motivate individuals to show prosocial behavior and repair the negative consequences of the action that caused these feelings, by engaging in behavior such as apologizing and repairing (Davidson et al., 2018; Drummond et al., 2017; Lewis, 1993; Muris, 2015; Muris & Meesters, 2014; Muris et al., 2016; Tangney, 1999; Tangney & Dearing, 2002). An observational study on social emotions in toddlerhood using a mishap task, the avoiders vs. amenders study, coded repairing behavior and telling the experimenter about the mishap as guilt-related behaviors (Barrett, Zahn-Waxler, & Cole, 1993). Another observational study on toddlers’ shame and guilt expressions using a mishap task had a likewise approach; telling the
experimenter or parent about the mishap, and repairing behavior were coded as related to guilt (Drummond et al., 2017).
Shame, on the other hand, involves a sense that the entire self is bad (Barrett, 2005; Drummond et al., 2017; Lewis, 2007; Tangney & Dearing, 2002). If the whole self is evaluated as being bad, instead of a specific action, there is no motivation for reparative behavior as there is little one can do to repair the whole self. Therefore withdrawing or hiding oneself from the situation and bad evaluations is the most useful reaction (Barrett, 2005; Drummond et al., 2017; Lewis, 1993, 2007; Tangney & Dearing, 2002). Shame is often considered as an inhibitor of prosocial behavior, as it leads to withdrawal, denial, and avoidance (Barrett, 2005; Tangney, 1999; Tangney et al., 1996). Lewis (1993) proposes that the physical reaction to shame is the shrinking of the body, as if to disappear from evaluation. Shame is expressed by withdrawing behavior—such as aversion of the eyes, head or the whole body—while having a neutral or negative facial expression (Barrett, 2005; Drummond et al., 2017; Lewis, 1993). The avoiders vs. amenders study coded latency to look at the experimenter and gaze, head, and body aversions as shame-related behaviors (Barrett et al.,
1993). The study by Drummond et al. (2017) coded gaze and bodily aversions as shame-related behaviors. However, all behaviors described above can also be rooted in fearful avoidance, and some researchers did not go as far to interpret these behaviors as indicators of shame, as this behavior could indicate general arousal or anxiety (Barrett, 2005; Hobson et al., 2006). Since it is also suggested that it may be too early for toddlers to experience shame as a distinct emotion (Tangney & Dearing, 2002), we refer to these behaviors as shame-like avoidance.
Embarrassment involves evaluations about the presented self, as opposed to the core self, in shame (Lewis, 1993; Tangney, 1999; Tangney et al., 1996). This generates more short-term and situation-specific negative evaluations about oneself (Tangney, 1999; Tangney et al., 1996). Apart from the prosocial function of guilt, expressions of embarrassment also serve an important social function; they are indicative of appeasement and serve as a non-verbal acknowledgement of shared social standards (Feinberg et al., 2012; Miller, 2007; Tangney, 1999). Miller (2007) describes a distinctive pattern of nonverbal behavior indicative of embarrassment: averted gaze while trying not to smile by compressing the lips, which usually fails and ends in a smile or grin. He stresses that not all of these actions always occur. Other researchers describe the expression of embarrassment just as withdrawing behavior accompanied by a smile (Barrett, 2005; Lewis, 1993, 1997, 2007). The avoiders vs. amenders study coded embarrassment as the presence of a smile followed by gaze aversion within 5 seconds (Barrett et al., 1993).
Social Emotions in Early Childhood
Social emotions require more complex cognitive and emotional abilities than basic emotions, such as happiness, sadness, or anger, and therefore appear later in children’s development than basic emotions (Lightfoot, Cole, & Cole, 2013; Muris & Meesters, 2014). Because children need to possess specific cognitive skills to enable them to experience social
emotions (Muris & Meesters, 2014), it has been discussed whether social emotions are already present in children before the age of 2 years (Barrett et al., 1993; Drummond et al., 2017). The emergence of self-awareness is assumed to be an important condition for experiencing social emotions (Davidson et al., 2018; Kochanska, Gross, Lin, & Nichols, 2002; Lewis, 2007; Muris & Meesters, 2014), which normally develops between 18 and 24 months (Barrett et al., 1993; Cole, Barrett, & Zahn-waxler, 1992; Drummond et al., 2017; Muris & Meesters, 2014; Shaffer, 2005). Lewis (1993) proposed a cognitive-attributional theory as explanation for the development of social emotions. This theory assumes that, next to self-awareness, children need to establish standards, rules, or goals, the evaluation of success or failure in regard to these, and the attribution of the self (Lewis, 1993). Lewis (2007) suggests that some social emotions which do not have evaluation of the self as their basis, only need self-awareness and thus emerge around the age of 2 years. Nonevaluative or exposure embarrassment, for example, is based on cognitions about the self, but is not based on evaluations of the self (Lewis, 2007). However, before children are able to realize they did something wrong, they need to incorporate social rules and norms, and additionally need to be aware that they are responsible for their own behavior (Davidson et al., 2018; Lewis, 1993, 2007; Muris & Meesters, 2014). Children are normally capable of doing so around the age of 3 years, resulting in the emergence of the more complex, evaluative social emotions such as guilt, evaluative embarrassment, and shame (Lewis, 2007). Another skill related to
experiencing social emotions is the awareness of and reflection on how other people evaluate the child’s behavior (Muris & Meesters, 2014). This skill may be related to the development of a ToM, which is the ability to understand others’ mental states and predict behaviors accordingly and which appears later in the development, around the age of 4 years (Derksen, Hunsche, Giroux, Connolly, & Bernstein, 2018).
theory and the necessity of a fully developed awareness of the self in relation to others (Draghi-Lorenz et al., 2001; Hobson, 1990; Reddy, 2005; Zinck, 2008). They acknowledge the need of an understanding of the self in relation to others, but propose that this ability is available at an earlier age and may even be innate (Draghi-Lorenz et al., 2001; Reddy, 2005). Additionally, it is stressed that this interpersonal awareness should be considered a continuous process rather than an ability that is achieved at a specific point in time, as suggested by studies that define self-awareness through passing a self-recognition in a mirror task (Draghi-Lorenz et al., 2001). Like-minded, Zinck (2008) proposed the presence of pre-forms of complex social emotions in early childhood that do not require full-fledged interpersonal awareness and ToM. Rather, an interaction between these pre-forms of social emotions and precursors of ToM are what later develops into fully developed interpersonal awareness and ToM, and may cause the ability to experience and express social emotions at an earlier age (Zinck, 2008). Several studies indeed found expressions of social emotions in early childhood. The study of Barrett (2005) showed that 17-month-old children displayed behaviors indicative of guilt, embarrassment, and avoidance after breaking the experimenters toy. Other
researchers found that 22-month-old children showed guilt-related behaviors after being led to believe they had damaged a valuable object of the experimenter (Kochanska et al., 2002). Additionally, a study with children ranging from 25 to 36 months found distinctive guilt- and shame-related behaviors after breaking the experimenters doll (Barrett et al., 1993).
Autistic Traits and Social Emotions
Children with high levels of autistic traits may be characterized by deficits in social emotions (Davidson et al., 2018, 2017). A study that compared 7- to 14-year-old children with and without ASD, found that children with ASD showed significantly less proneness to guilt than typically developing children, based on self-reports (Davidson et al., 2018). A multi-informant study compared 4- to 18-year-old children with and without ASD, based on parent,
teacher, and child reports (Muris et al., 2016). They found that children and adolescents with ASD showed significantly less guilt than children and adolescents without ASD. Another study, based on self-reports, found that adults with clinical levels of autistic traits showed significantly less proneness to guilt than adults without autistic traits (Davidson et al., 2017). Hobson et al. (2006) compared 12 6- to 12-year-old children with ASD to 12 matched non-ASD children. They found that the children with non-ASD experience, as reported by their parents, less guilt and embarrassment. Additionally, they found that during a guilt-eliciting mishap task, only one of the children with ASD showed signs of guilt, whereas ten of the non-ASD children showed signs of guilt. Hobson et al. (2006) also compared 20 6- to 16-year-olds with ASD to 19 matched non-ASD children during an embarrassment-eliciting task. They found that none of the children with ASD showed gaze aversion accompanied by a smile—a sign of embarrassment or shyness, whereas over 60% of the children without ASD showed these embarrassment-related behaviors. In addition, a study comparing pre-school children with ASD, Down syndrome, and typically developing children during a mirror task, found that children with ASD had an absence of coy-smiles, unlike the other children in the study (Reddy et al., 2010). On the other hand, for shame—and more specifically, shame-like avoidance—studies found mostly positive associations with ASD. The multi-informant study by Muris et al. (2016) found evidence for higher levels of shame in children and adolescents with ASD compared to children and adolescents without ASD. Additionally, adults with clinical levels of autistic traits reported significantly more shame proneness than adults without autistic traits (Davidson et al., 2017). A study compared 2- to 14-year-old children with and without ASD and found that children with ASD showed significantly more avoidant symptoms than typically developing children, as reported by their parent (Davis et al., 2011). Higher levels of shame-like avoidance correspond with the social impairments related to autism, which may cause the prevailing response in social situations to be avoiding and
withdrawing (Muris et al., 2016). Less proneness to guilt and embarrassment in children with higher levels of autistic traits may be caused by more difficulties in recognizing social norms (Muris et al., 2016), and recognizing and understanding emotions and other mental states of others (Harms et al., 2010; Schaller & Rauh, 2017). It has been shown that children diagnosed with ASD may show social emotions in situations where there was explicitly explained that the child was violating norms (Kasari, Chamberlain, & Bauminger, 2001), which suggests that children with ASD are able to express social emotions, but have difficulties
understanding when they violated a norm and how it may influence others. Hence, the dysregulations of social emotions in children with high levels of autistic traits may partly be due to the deficits in socio-cognitive abilities, such as ToM.
Socio-Cognitive Abilities
Social cognition concerns the cognitive mechanisms underlying our social behavior, that is, how individuals perceive, implement, and evaluate information gathered in social situations (Schaller & Rauh, 2017; Senju, 2013). It emerges gradually during life (Sirois & Jackson, 2007). ToM, which is an important socio-cognitive ability, may be an underlying cause for the behavioral peculiarities in individuals with ASD (Buitelaar, Van Der Wees, Swaab-Barneveld, & Van Der Gaag, 1999). ToM represents the ability to attribute mental states—feelings, thoughts and intentions—to one-self and others and the ability to understand and explain behavior of others based on their mental states (Mazza et al., 2017; Premack & Woodruff, 1978).
False belief, the realization that others can have diverging beliefs and intentions, is an important aspect in the development of ToM, and normally develops around the age of 4 years (Mazza et al., 2017; Poulin-Dubois & Yott, 2018). There are, however, early precursors or indicators of ToM that support the emergence of socio-cognitive abilities, which already develop before the age of 2 years (Sirois & Jackson, 2007). Another important component of
social cognition is the recognition and understanding of emotions (Denham, 1986; Denham, Zoller, & Couchoud, 1994). Emotion recognition—the ability to accurately assess emotions based on external cues, such as facial expressions—which is necessary to take the emotional perspective of others (Buitelaar et al., 1999), may thus also be of importance for children’s socio-cognitive abilities. The first manifestations of emotion recognition start to develop around the age of 7 to 10 months (Leppänen & Nelson, 2006; Shaffer, 2005). Around the age of 3 years a more complex understanding of emotions develops, that enables children to understand that certain situations generally elicit certain emotions (Lane, Wellman, Olson, LaBounty, & Kerr, 2010). Knowledge about others’ emotions is important to develop because it enables us to deduce how someone is feeling and adjust our behavior accordingly (Buitelaar et al., 1999; Shaffer, 2005). Deficits in emotion recognition could result in impairments in social interactions and social relations (Buitelaar et al., 1999; Hobson, Ouston, & Lee, 1988).
Children with autism usually experience impairments in socio-cognitive abilities (Baron-Cohen, Leslie, & Frith, 1985; Buitelaar et al., 1999; Burnside, Wright, & Poulin-Dubois, 2017; Harms et al., 2010; Heerey, Keltner, & Capps, 2003b; Mazza et al., 2017; Misailidi, 2018; Muris & Meesters, 2014; Schaller & Rauh, 2017; Senju, 2013). A meta-analysis of functional neuroimaging studies comparing children and adults with and without ASD, revealed hypofunction in brain regions associated with socio-cognitive abilities—i.e., the right anterior insula—during social tasks in individuals with ASD (Martino et al., 2009). It has been found, by using the adapted belief-based anticipatory looking task (Thoermer,
Sodian, Vuori, Perst, & Kristen, 2012), that preschoolers with ASD showed lower levels of implicit false belief than typically developing preschoolers (Burnside et al., 2017). Also, adolescents with ASD seemed to score lower on emotion recognition and first and second order false belief, using the Animated-Theory-of-Mind test, than their typically developing peers (Schaller & Rauh, 2017). Burnette et al. (2005) compared first- and second-order belief
of children (Mage = 11 years) with and without ASD by using the M&Ms false belief task
(Perner, Frith, Leslie, & Leekam, 1989) and the second-order ToM task (Baron-Cohen, 1989). They found that children with ASD scored significantly lower on both the first- and second-order false belief tasks than the children without ASD (Burnette et al., 2005). A possible explanation for the deficits in socio-cognitive abilities may be that children with autistic traits often experience reduced social input because of impairments in social communication and interaction, which may result in a deprived development of socio-cognitive abilities (Burnside et al., 2017; Leppänen & Nelson, 2006). In addition, less social input combined with a lack of orientation to faces, which is common in children with high levels of autistic traits, causes less exposure to various facial expressions, which may lead to difficulties in emotion recognition (Harms et al., 2010).
Impaired socio-cognitive abilities could in turn cause dysregulated social emotions. Being able to form beliefs about the ways other people evaluate your behavior, can contribute to experiencing social emotions. For example; individuals will feel more guilt when they believe that others believe their action was wrong (Davidson et al., 2018; Heerey et al., 2003b; Misailidi, 2018; Muris & Meesters, 2014). Thus, children need to have developed some socio-cognitive skills such as perspective taking. ToM and emotion recognition can help children understand what others may think and feel, and thus, may promote adaptive social emotions such as guilt and embarrassment (Denham et al., 1994; Heerey et al., 2003b; Misailidi, 2018). Impairments in ToM could also cause failure to understand social
transgressions, which in turn could lead to less guilt or embarrassment; when individuals are not aware of violating a norm or rule, they will not believe that others believe their action was wrong and thus, will not show guilt or embarrassment (Heerey et al., 2003b; Muris et al., 2016). When explicitly explained to a child with ASD that a norm or rule is violated, which eliminates the need for ToM, children with ASD can show social emotions (Kasari et al.,
2001).
As socio-cognitive abilities involve using others’ mental states to predict their future behavior (Burnette et al., 2005), several scholars suggested that a lack of reasoning about others’ emotions and intentions could cause a general coping strategy of avoidance in difficult social situations to avoid distress caused by high levels of arousal (Hezel & Mcnally, 2014; Mirenda, Donnellan, & Yoder, 1983). Therefore, children with higher levels of autistic traits may show more shame-like avoidance in social situations, as a result of impaired socio-cognitive abilities. Kuusikko et al. (2008) compared 8 to 15 year old children with and without ASD on their levels of social anxiety and social avoidance. They found that children with ASD, as reported by their parents, showed more withdrawal from social situations. Additionally, self-reports of the children showed that children with ASD above 12 years old, showed more social and behavioral avoidance than the control group (Kuusikko et al., 2008). They suggest that in children with ASD, deficits in the development of cognitive-emotional skills and self-awareness, resulting in impaired social interaction skills and increased arousal in interpersonal interactions, may lead to increased avoidance from social situations. Thus, deficits in socio-cognitive abilities may play a role in reduced expressions of guilt and embarrassment, and elevated levels of shame-like avoidance in children with autistic traits. It is, however, important to stress that it remains unclear whether this avoiding behavior truly has shame as underlying emotion, or it originates simply from increased arousal and fear in social situations.
Present Study
This study investigated whether levels of autistic traits are associated with the expressions of guilt, embarrassment, and shame-like avoidance after a misbehavior, such as breaking a favorite toy of the experimenter in children aged 2 to 5 years old. It was also assessed whether these associations were mediated by socio-cognitive abilities. It was
hypothesized that children with higher levels of autistic traits would show less guilt (Davidson et al., 2017; Hobson et al., 2006; Muris et al., 2016) and embarrassment (Hobson et al., 2006), and more shame-like avoidance (Davidson et al., 2017; Davis et al., 2011; Muris et al., 2016). Children with higher levels of autistic traits were also expected to have lower levels of socio-cognitive abilities (Baron-Cohen et al., 1985; Burnside et al., 2017; Heerey, Keltner, & Capps, 2003a; Muris & Meesters, 2014). Finally, it was hypothesized that the relation
between autistic traits and social emotions was mediated by socio-cognitive abilities (Heerey et al., 2003a; Kuusikko et al., 2008; Muris et al., 2016) in the way that higher levels of autistic traits would be associated with lower levels of socio-cognitive abilities, which, in turn, would relate to less expressed guilt and embarrassment, and more shame-like avoidance.
Method Participants
Participants of this study were 98 2- to 5-year-old children (50% girls, M = 4.06 years, SD = 1.13 years, 92% Caucasian), accompanied by one of their parents, aged 22 to 48 years (83% mothers, M = 35.76 years, SD = 6.14 years). They visited the Family laboratory at the University of Amsterdam. Parents mostly had a Dutch ethnicity (81.6%), were relatively highly educated (11% doctorate, 30% university degree, 35% college degree, 24% high school/vocational degree) and were, based on monthly household income, from various socio-economic status (42% < €4000, 31% €4000-€6000, 22% > €6000).
The children and their parents were recruited by letters that were sent out via primary schools, daycare centers and pre-schools, by handing out information letters in front of crowded public areas frequently visited by families with little children, such as parks, and children museums in Amsterdam, and through online advertising on social media. Parents filled in an informed consent form to register for this study, but were free to drop out anytime, without giving any reason. The study has been approved by the ethics committee of the Research Institute of Child Development and Education of the University of Amsterdam (n. 2018-CDE-8675).
Children’s Levels of ASD Traits Reported by Parent
The Social Responsiveness scale – Preschool Version (SRS; Constantino & Gruber, 2005; Pine, Luby, Abbacchi, & Constantino, 2006) was used to measure the level of ASD traits. The SRS is a parent report to measure children’s autistic traits which consists of 65 items with a four point rating scale ranging from not true to always true, regarding a child’s ability to engage in an emotionally appropriate way in interactions with others (Constantino et al., 2003). An example item is: “My child avoids starting social interactions with peers or adults”. The SRS has five subscales: social awareness, social cognition, social
communication, social motivation and autistic mannerisms (Constantino et al., 2003). Most items were indicative of behaviors regarding autistic traits, so higher scores represented higher levels of autistic traits, but some items were reversed and so, higher scores represented lower levels of autistic traits. Values for these items were reversed, so higher scores indicated higher levels of autistic traits. Afterwards, scores on all items were averaged which resulted in the final autistic traits score. The test-retest reliability of the SRS measured by Constantino et al. (2003) was 0.83. Internal consistency was excellent in previous studies, α = .93
(Constantino & Gruber, 2005), α = .97 (Bölte, Poustka, & Constantino, 2008),and internal consistency for the SRS found in the present study was good, α = .82. The preschool version only differs from the SRS on the basis of developmental appropriateness of the wording to describe behaviors of children in the respective age group, and analysis revealed no effects of test version (Pine et al., 2006).
Measurement and Coding of Social Emotions
The Broken Toy mishap (Barrett et al., 1993; Cole et al., 1992; Kochanska, Casey, & Fukumoto, 1995) adapted based on Drummond et al. (2017) was used to observe social emotions. During the Family lab visit, the parent received an instruction letter to inform them about the task, and the task was only performed with additional permission from the parent. Parents were instructed to remain as neutral las possible during this task. The experimenter (henceforth called “E”) presented a teddy bear (henceforth called “Teddy”) to the child with emphasis on the emotional value of Teddy for E by mentioning it was E’s favorite childhood toy. Teddy was then attached to the wall with a Velcro patch and the child was told that he/she was allowed to play with Teddy while E was gone. Teddy was rigged in the way that the arm and/or leg fell off when the child started playing with it. Around two minutes after the child broke Teddy, E returned to the room following five cues with 15 seconds in between, in which the child got stimulated to respond to the situation. The cues were: 1. loudly saying:
“Alright, thanks a lot and see you soon!” just outside the room, so the child knew E was returning to the room. 2. Looking at the broken bear with a neutral facial expression. 3. “What happened to Teddy?”. 4. “What happened so his arm/leg fell of?”. 5. “Teddy was my favorite bear”. After the five cues the child got debriefed by saying: “Oops, I forgot that Teddy was already broken! I can fix him, I will be right back with a repaired Teddy.”. After leaving the room, E returned with an identical, not broken Teddy and said: “See, he is as good as new. Now Teddy is happy again, I am happy again and you can be happy again as well.”. If the child showed severe signs of distress (e.g., looking like they would start crying) at any point during the task, E immediately debriefed the child regardless of the other cues that had to come. The Broken Toy mishap had a mean duration of 154.91 seconds (SD = 49.04).
Afterwards, different behaviors were coded using the video records of the task and the coding program ‘The Observer XT 13’ (Zimmerman, Bolhuis, Willemsen, Meyer, & Noldus, 2009). The expressions of the social emotions were coded based on coding systems of Barrett (2005) and Drummond et al. (2017), which was adjusted to account for different indicators of social emotions used also in other studies (e.g., Vaish, Carpenter, & Tomasello, 2016). Descriptions and examples of coded behaviors are displayed in Table 1. Coded behaviors were not mutually exclusive, so different behaviors could happen at the same time. Gaze, head, and body aversion, smiling, repairing Teddy, and latency to talk about Teddy were coded as state events, so durations of these behaviors were coded and could be converted to proportions, relative to the total duration of the mishap task. Aversion was coded only when E was in the room, and thus proportions were calculated relative to the total duration that E was in the room during the mishap. Frequencies were coded for the amount of verbal expressions. The behaviors were coded by trained master students and Inter-rater reliability was
established on 12 observations (18%). Cohen’s kappa corrected for kappa max was κ = .97 for gaze aversion, κ = .99 for head aversion, κ = .94 for body aversion, κ = .57 for gaze aversion
during smile, κ = .52 for head aversion during smile, κ = .97 for not talking about Teddy, κ = .90 for talking about Teddy, κ = .99 for repairing Teddy.
Table 1
Coded Behaviors with Description and Example
As indicated earlier, guilt has been established to be associated with repairing behavior (e.g., Tangney, 1999). Also, verbal expressions that were comforting and showed concern of what would happen are expected to reflect guilt (Barrett et al., 1993; Ioannou et al., 2013). Embarrassment has been widely associated with aversions during smiling (Barrett, 2005; Tangney et al., 1996). Shame is typically expressed by avoiding behavior while having a
Behavior Description Example
Gaze aversion Averted gaze from E The child looks away (e.g., at the floor) after looking at E
Head aversion Averted head from E The child averts or hides their head (e.g., cover face with hands) after facing E
Body aversion Averted body from E The child averts or hides their body after facing E
Lips Indicating a smile or no smile Lip corners up; indicating smile. Lip corners neutral or down; not indicating smile
Repairing Repairing Teddy or not repairing Teddy
The child is trying to put the arm, leg, or stuffing back
Talking Talking about Teddy or not talking about Teddy to parent and E (once E entered the room)
“Look, Teddy is broken”
Verbal expressions Taking responsibility “I made the arm/leg fall of” Acknowledging feelings/taking
perspective of E
“She (E) is going to be sad”
Comforting E “I can fix it”
Concern what will happen to E or Teddy
neutral or negative facial expression (Barrett, 2005; Drummond et al., 2017; Tangney et al., 1996). However, some researchers pointed out that avoidance behaviours do not necessarily have shame as underlying emotion, as these behaviors could also indicate general inhibition or anxiety (Barrett, 2005; Hobson et al., 2006). Therefore, we named these behaviours (aversions, latency) shame-like avoidance. A principal component analysis using Oblimin rotation, which allows for factors to correlate, was performed to find which coding behaviors loaded onto the same factor, underlying a specific social emotion. Factor loadings are
displayed in Table 2. Based on the screeplot, four latent factors were used, as they had eigenvalues larger than 1 (Kaiser criterion; Kaiser, 1960). Together, the factors explained 71% of the variance. The factor loadings indicated that the children’s responses to the Broken Toy mishap were organized in four patterns that are consistent with theory, and similar to previous studies (Barrett, 2005; Drummond et al., 2017). However, they showed a distinction between physical shame-like avoidance: gaze, head, and body aversion, and verbal shame-like avoidance: latency to speak to their parent and to E once E enters the room during the mishap, and the inverted score for the amount verbal expressions that showed perspective- and
responsibility taking. An overview of the social emotions derived from the coded behaviors is displayed in Table 3.
Table 2
Pattern Matrix of Principal Component Analysis
Table 3
Outcome Variables Derived from the Coded Behaviors
Measurement of Socio-Cognitive Abilities
Both ToM and emotion recognition were measured. A shortened version of the Theory of Mind-test-Revised (ToM-test-R; Steerneman & Meesters, 2009) was used; a structured interview with 14 short stories about which the child had to answer 36 questions for measuring ToM in children. The ToM-test-R has three subscales: ToM 1, which measured precursors of ToM such as emotion understanding and pretense, ToM 2, which measured the
Behaviors Components
Guilt Embarrassment Physical
avoidance
Verbal avoidance Proportion repairing .605
Verbal: Comforting E .866 Verbal: Expressing concern .836
Proportion gaze aversion (smile) -.972 Proportion head aversion (smile) -.979
Proportion gaze aversion .899
Proportion head aversion .916
Proportion body aversion .773
Latency to talk .777
Verbal: Taking responsibility -.663
Verbal: Taking perspective -.659
Variable Composed of
Guilt Proportion of repairing behavior and amount of verbal expressions of guilt
Embarrassment Gaze, body, or head aversion while smiling
Physical shame-like avoidance Gaze, body, or head aversion while having lip corners in neutral or downward position
Verbal shame-like avoidance Latency to start talking about Teddy to the parent and T, and inverted score for amount of verbal expressions
first manifestations of true ToM such as false belief and ToM 3, which measured more
advanced ToM aspects (Ahmadi, Jalaie, & Ashayeri, 2015). The ToM-test-R is normally used for children aged 4 to 12 years old. In this study only items 1, 3, 4 and 5 were used,
representing only ToM 1 and 2, because the following tasks were too advanced for children aged two to five years. These items measured four different milestones in the development of ToM; item 1 measured pretense, item 3 emotion understanding and first order belief, and item 4 and 5 false belief. Pretense is the ability to pretend, whereas first order belief refers to the understanding of someone’s actions and false belief to the understanding that someone can have different beliefs than reality and act accordingly (Leslie, 1987). Each ToM-test item was scored as either ‘good’ or ‘false’ and got a score of respectively 1 or 0, where a higher score reflected more developed ToM. Pretense and emotion recognition represented ToM 1 and first order belief and false belief represented ToM 2. Internal consistency of the complete ToM-test-R in a previous study was excellent, α = .92 (Ahmadi et al., 2015), internal consistency in the present study was good for ToM 1, α = .89, and good for ToM 2, α = .81.
An adapted version (Pears & Fisher, 2005) of the emotion understanding task
(Denham, 1986) was used to measure emotion recognition; a task using four pictures of a boy or girl, matched to the gender of the child, showing happy, sad, scared and angry emotions. The task was divided into two subscales: expressive emotion recognition and receptive emotion recognition (Pears & Fisher, 2005). While expressive emotion recognition was measured by presenting the pictures one-by-one and asking which picture represents which emotion, receptive emotion recognition was measured by presenting all four pictures at the same time and asking the child to point to the requested emotion (Pears & Fisher, 2005). A correct answer was scored as 1 and a wrong answer as 0, resulting in a sum score for emotion recognition based on both receptive and expressive emotion recognition tasks, where a higher score represented higher levels of emotion recognition. Internal consistency for the emotion
understanding task found in earlier studies was good, α = .84 (Pears & Fisher, 2005), internal consistency for the expressive emotion understanding task in the present study was
unacceptable, α = .31, and for the receptive emotion recognition task was acceptable, α = .74. Therefore, we only used the results of the receptive emotion recognition task in this study.
Measurement of Verbal Ability
Children’s verbal ability was measured using the Peabody Picture Vocabulary Test – Third Edition (Dunn & Dunn, 1997). Children were shown a sheet with four pictures and they were asked to point to the correct picture corresponding the called object. For this study we used the first nine sets, with each 12 objects. This procedure was continued until the child pointed to nine or more incorrect objects in one set. Then a total score was computed by counting the total amount of objects and subtracting the number of incorrect answers. This total score was then transformed into a normalized score that displayed children’s vocabulary level relative to their age.
Data Analyses
The data was checked for outliers (± 3 SD) and the distributions of the residuals were checked on normality. Outliers were Winsorized by changing the value to the closest value within the ± 3 SD range (Wilcox, 2005). Due to procedural or video recording errors, and the need for parental consent for the Broken Toy mishap, 32 observations (33%) of the social emotions were missing. Due to procedural or video recording errors, children’s insufficient speaking or understanding of Dutch, or unwillingness or lack of time to fulfil the task or questionnaire, 4% of the autistic traits questionnaire, 11% of the socio-cognitive ability tasks, 7% of the reported children’s age, and 54% of the verbal ability task were missing. Analysis of missing data showed a total of 22.76% missing values, and Little’s MCAR test indicated that the data is missing completely at random, χ2(104) = 124.29, p = .085. Also, missingness
samples t-test, all p values > .085. Missing values were estimated using full information maximum likelihood estimation, and, subsequently, imputed with regression imputation in Amos (Arbuckle, 2016; Version 24). The imputed dataset was used in the analyses.
To test the hypotheses about autistic traits, socio-cognitive abilities, and social emotions, we used path analyses which were carried out in Amos (Arbuckle, 2016; Version 24). In one model, we tested (1) whether autistic traits predicted expressions of social
emotions directly; (2) whether there was a partial mediation of socio-cognitive abilities in the relation between autistic traits and social emotions; (3) whether these direct and indirect effects were dependent on levels of child age and child verbal abilities. Thus, the model consisted of autistic traits as predictor, socio-cognitive abilities as a mediator, guilt, embarrassment, verbal shame-like avoidance, and physical shame-like avoidance as
outcomes, and child age and normed verbal ability as covariates. Based on our hypotheses, we estimated paths from the predictor to all outcomes, directly and through the mediator. Since the principal component analysis showed that the four factors did not correlate highly, all r values < .08, we did not correlate the error terms of the four outcome variables in our model, resulting in six degrees of freedom. For all significance tests, an alpha level of α = .050 was used. We also used bootstrapping on 1000 samples to obtain the bias-corrected bootstrapped 95% confidence intervals. Overall model fit was inspected based on the χ2 test, in which a significant χ2 value indicates a significant deviation between the model implied, and sample
covariance matrix, meaning that the model does not fit the data well. Next to the χ2 statistic,
the Comparative Fit Index (CFI) and the Root Mean Square Error of Approximation (RMSEA) were interpreted. CFI values of larger than .95 were considered as indicative of good fit (Hu & Bentler, 1999). RMSEA values smaller than .05 indicate close fit, values smaller than .08 indicate reasonable fit, and smaller than .10 mediocre fit (Browne & Cudeck, 1992). However, as RMSEA values tend to be erratic with small sample sizes (n < 100), the
90% confidence interval, which gives an indication of the imprecision of the RMSEA value, will also be inspected (Chen, Curran, Bollen, Kirby, & Paxton, 2008). Direct, indirect, and total effects were tested by means of bootstrapped and bias-corrected 95% confidence intervals.
Results Preliminary Results
Six outliers above the range of 3 SD were Winsorized (Wilcox, 2005), two for latency to start talking about Teddy, one for embarrassment, and two for guilt. Despite adjusting the outliers, the residuals of these variables were not normally distributed. To improve the distributions of the residuals and reduce the skewness, guilt was square root transformed, skewnessguilt = 0.60 (SE = 0.30), and latency to start talking about Teddy was e logarithm
transformed, skewnesslatency = 0.58 (SE = 0.30). As embarrassment was still skewed after
transformations, the winsorized variable was used in the analyses, skewnessembarrassment = 2.10
(SE = 0.30).
According to a one sample t-test, the average total score of autistic traits in the current sample (M = 36.46, SD = 15.31), differed significantly with the average of the normative population in the validation study (M = 33.00, SD = 14.90), t(94) = 2.19, p < .050. However, the average of the current sample was still quite below the clinical cutoff of 60 (Aldridge, Gibbs, Schmidhofer, & Williams, 2012). A study assessing the prevalence of parent-reported ASD in children is found to be 1.7% (Russell, Rodgers, Ukoumunne, & Ford, 2014). In the current sample, 9% (n = 9) of the children had clinical levels of autistic traits, which means the prevalence of clinical levels of autistic traits in this study was higher than in the general population. According to an independent sample’s t-test, there were no significant differences between boys and girls on autistic traits, socio-cognitive abilities, or the social emotions, all p values > .101. Children’s sex was therefore not included in the analyses. Descriptive statistics
for all the variables in our model are presented in Table 4. Pearson’s correlations between children’s autistic traits, socio-cognitive abilities, and social emotions are displayed in Table 5. Autistic traits had a significant negative correlation with socio-cognitive abilities,
embarrassment, and normed verbal ability, and a significant positive correlation with verbal shame-like avoidance. Socio-cognitive abilities had a significant positive correlation with guilt, embarrassment, and child age. Guilt had a significant positive correlation with child age and normed verbal ability. Verbal shame-like avoidance had a significant negative correlation with normed verbal ability. As there were significant associations with children’s age and verbal ability relative to age, these were added to the model to control for their effects. The inspected model is displayed in Figure 1. Based on the χ2 test of overall model fit, the null
hypothesis of perfect fit to the data could not be rejected, χ2(6) = 5.90, p = .434, CFI = 1.00, which indicated good fit, RMSEA = 0.00, which indicated close fit, with 90% CI [0.00, 0.13]. Thus, the model fitted the data well and was not adjusted.
Table 4
Descriptive Statistics of Children’s Autistic Traits, Socio-Cognitive Abilities, Social Emotions, Age, and Normed Verbal Ability.
N Mean (SD) Range
Autistic traits 98 1.60 (0.23) 1.16 – 2.33
Socio-cognitive abilities 98 10.24 (4.81) 1.00 – 19.00
Guilta 98 0.98 (0.28) 0.08 – 1.86
Proportion repairing 66 0.23 (0.21) 0.00 – 0.86
Verbal: Comforting experimenter 58 0.12 (0.46) 0.00 – 3.00 Verbal: Expressing concern 58 0.16 (0.62) 0.00 – 3.00
Embarrassment 98 0.01 (0.01) 0.00 – 0.07
Proportion gaze aversion (smile) 66 0.01 (0.02) 0.00 – 0.12 Proportion head aversion (smile) 66 0.01 (0.02) 0.00 – 0.14
Verbal shame-like avoidancea 98 0.04 (0.74) -1.74 – 2.11
Latency to talk 66 0.90 (0.46) 0.16– 1.88
Verbal: Taking responsibility 58 0.64 (1.02) 0.00 – 4.00 Verbal: Taking perspective 58 0.17 (0.43) 0.00 – 2.00
Physical shame-like avoidance 98 0.52 (0.21) 0.04 – 0.98
Proportion gaze aversion 66 0.65 (0.22) 0.11 – 1.00
Proportion head aversion 66 0.52 (0.23) 0.01 – 0.98
Proportion body aversion 66 0.37 (0.33) 0.00 – 1.00
Child age (months) 98 48.31 (13.31) 24.00 – 72.00
Normed verbal ability 98 106.15 (9.96) 72.00 – 132.00
Note. a. variables were standardized and averaged into a composite score.
Table 5
Pearson’s Correlation Coefficients for Children’s Autistic Traits, Socio-Cognitive Abilities, Social Emotions, Child Age in Months and Normed Verbal Ability.
1 2 3 4 5 6 7
1. Autistic traits –
2. Socio-cognitive abilities -.29** –
3. Guilt -.15 .54** –
4. Embarrassment -.25* .29** -.04 –
5. Verbal shame-like avoidance .31* -.14 -.14 -.03 –
6. Physical shame-like avoidance .17 -.02 .08 .02 .03 –
7. Child age -.16 .79** .56** .14 .03 -.11 –
8. Normed verbal ability -.46** .18 .53** -.12 -.23* -.08 .12
Note. Significant correlations are highlighted with asterisks representing significance levels: *p < .050, **p <
Relation between Autistic Traits and Social Emotions through Socio-Cognitive Abilities
In order to test our hypothesized model, we tested a model (Figure 1) with autistic traits as the predictor—along with child age and normed verbal ability as covariates—socio-cognitive abilities as mediator, and guilt, embarrassment, verbal shame-like avoidance, and physical shame-like avoidance as the outcomes. We inspected the direct effects for the relation between the variables while taking all other effects into account, and the indirect and total effects to establish whether there was mediation by socio-cognitive abilities.
Standardized direct effects are displayed in Figure 1, and standardized indirect and total effects are presented in Figure 2. The unstandardized regression coefficients and their bootstrapped and bias-corrected confidence intervals are displayed in Table 6.
Autistic traits. Normed verbal ability significantly and negatively related to levels of
autistic traits, indicating that children with higher levels of autistic traits had less-developed normed verbal abilities. Levels of autistic traits did not relate to child age, indicating that there were no differences in levels of autistic traits based on children’s age.
Socio-cognitive abilities. Autistic traits significantly and negatively related to
socio-cognitive abilities, indicating that children with higher levels of autistic traits had lower levels of cognitive abilities. Children’s age significantly and positively related to
socio-cognitive abilities, indicating that older children had better-developed socio-socio-cognitive abilities. Socio-cognitive abilities did not significantly relate to normed verbal abilities, indicating that there were no differences in socio-cognitive abilities based on children’s normed verbal ability. These three variables explained 66% of the total variance in socio-cognitive abilities.
Guilt. Autistic traits significantly and positively related to guilt directly, indicating
that children with higher levels of autistic traits showed more guilt. Guilt was not significantly related to socio-cognitive abilities. Children’s age significantly and positively related to guilt,
indicating that older children showed more guilt. Normed verbal ability significantly and positively related to guilt, indicating that children with more verbal abilities relative to their age showed more guilt. The predictors in the model explained 58% of the total variance in guilt.
Next to the positive direct effect of autistic traits on guilt, we found a significant negative indirect effect of autistic traits on guilt. This indicates that for children with equally developed socio-cognitive abilities, children with higher levels of autistic traits displayed more guilt. However, when taking into account the effect of socio-cognitive abilities, children with higher levels of autistic traits have less developed socio-cognitive abilities, which, in turn, makes these children display less guilt. The total effect of autistic traits and socio-cognitive abilities on guilt was not significant. This is likely because the indirect and direct effect are of opposite signs thereby neutralizing each other, the case called inconsistent mediation (MacKinnon, Krull, & Lockwood, 2000).
Embarrassment. Autistic traits significantly and negatively related to embarrassment,
indicating that children with higher levels of autistic traits showed less embarrassment. Socio-cognitive abilities significantly and positively related to embarrassment, indicating that
children with better-developed socio-cognitive abilities showed more embarrassment. Normed verbal ability significantly and negatively related to embarrassment, indicating that children with more verbal abilities relative to their age showed less embarrassment. Embarrassment was not significantly related to child age. The predictors in the model explained 20% of the total variance in embarrassment.
We found a significant partial mediation by socio-cognitive abilities on the relation between autistic traits and embarrassment. Next to the significant negative direct effect of autistic traits on embarrassment, we found a significant negative indirect effect of autistic traits on embarrassment. This implies that children with higher levels of autistic traits tend to
show lower levels of socio-cognitive skills, and in turn, display less embarrassment. The significant direct effect of autistic traits on embarrassment did, however, also imply that there are mechanisms other than socio-cognitive abilities that explain less displays of
embarrassment in children with higher levels of autistic traits.
Verbal shame-like avoidance. Autistic traits significantly and positively related to
verbal avoidance, indicating that children with higher levels of autistic traits displayed more verbal avoidance. Socio-cognitive abilities did not significantly relate to verbal shame-like avoidance, indicating that children did not differ in the amount of verbal shame-like avoidance based on the levels of socio-cognitive abilities. Child age significantly and
positively related to verbal shame-like avoidance, indicating that older children showed more verbal shame-like avoidance. Verbal shame-like avoidance did not significantly relate to normed verbal abilities. The predictors in the model explained 14% of the total variance in verbal shame-like avoidance.
No significant indirect effect was found for the relation between autistic traits and verbal shame-like avoidance mediated by socio-cognitive abilities. However, both the direct effect of socio-cognitive abilities on verbal shame-like avoidance and the indirect effect showed a trend toward higher levels of autistic traits being related to more verbal shame-like avoidance through less socio-cognitive abilities. When direct and indirect effects were combined in the total effect of autistic traits on verbal shame-like avoidance, the effect was significant and positive, indicating that higher level of autistic traits were related to more verbal shame-like avoidance through the deficits in socio-cognitive abilities and other mechanisms combined.
Physical shame-like avoidance. Both autistic traits and socio-cognitive abilities did
not significantly relate to physical shame-like avoidance, indicating that children did not differ in the amount of physical shame-like avoidance based on levels of autistic traits or
socio-cognitive abilities. Physical shame-like avoidance was also not significantly related to child age or normed verbal abilities. The predictors in the model explained 6% of the total variance in physical shame-like avoidance.
For physical shame-like avoidance, no significant mediation effect of socio-cognitive abilities was found. All direct, indirect, and total effects were not significant. This indicated that levels of autistic traits were not related to the amount of displayed physical shame-like avoidance, and that there was no mediation of socio-cognitive abilities.
Table 6
Path Analyses with Autistic Traits as Predictor, Socio-Cognitive Abilities as Mediator, Social Emotions as Outcome, and Child Age and Verbal Ability Relative to Age as Covariates
Autistic traits Socio-cognitive abilities Estimate (SE) p [bootstrapped
95% CI] Estimate (SE) p [bootstrapped 95% CI] Socio-cognitive abilities Direct -3.47 (1.39) .012 [-6.62, -0.50] - - - Guilt Direct 0.27 (0.09) .004 [0.04, 0.45] 0.01 (0.01) .056 [0.00, 0.03] Indirect -0.04 (0.02) .028 [-0.13, -0.00] - - - Total 0.22 (0.09) .052 [-0.00, 0.41] - - - Embarrassment Direct -0.02 (0.01) .005 [-0.03, -0.00] .00 (0.00) .006 [0.00, 0.00] Indirect -0.00 (0.04) .014 [-0.01, -0.00] - - - Total -0.02 (0.11) .011 [-0.04, -0.01] - - -
Verbal shame-like avoidance
Direct 0.70 (0.35) .044 [-0.11, 1.31] -0.05 (0.03) .067 [-0.11, 0.02]
Indirect 0.16 (0.04) .075 [-0.02, 0.47] - - -
Total 0.85 (0.11) .021 [0.11, 1.42] - - -
Physical shame-like avoidance
Direct 0.76 (0.41) .060 [-0.46, 1.63] 0.04 (0.03) .166 [-0.02, 0.10]
Indirect -0.14 (0.04) .134 [-0.56, 0.03] - - -
Total 0.62 (0.14) .279 [-0.52, 1.48] - - -
Child age Normed verbal ability
Estimate (SE) p [bootstrapped 95% CI] Estimate (SE) p [bootstrapped 95% CI] Autistic Traits Direct -0.00 (0.00) .250 [-1.06, 0.18] -1.06 (0.26) .000 [-1.87, -0.54] Socio-cognitive abilities Direct 0.28 (0.02) .000 [0.23, 0.32] 0.01 (0.03) .833 [-0.07, 0.07] Guilt Direct 0.01 (0.00) .000 [0.00, 0.01] 0.02 (0.00) .000 [0.01, 0.02] Embarrassment Direct 0.00 (0.00) .153 [0.00, 0.00] 0.00 (0.00) .003 [-0.00, -0.00]
Verbal shame-like avoidance
Direct 0.02 (0.01) .049 [-0.00, 0.04] -0.01 (0.01) .318 [-0.03, 0.01]
Physical shame-like avoidance
Direct -0.02 (0.01) .120 [-0.04, 0.00] 0.00 (0.01) .994 [-0.02, 0.02]
Normed verbal ability
Figure 1. Partial mediation model with standardized direct effects.
Note. Black lines indicate significant paths, grey lines are non-significant paths. Dotted lines indicate the effects
of the covariates.
Figure 2. Standardized indirect and total effects of autistic traits on the social emotions through socio-cognitive abilities.
Note. Significant correlations are highlighted with asterisks representing significance levels: *p < .050, **p <
Discussion
The present study was unique in investigating the relation between autistic traits and expressions of social emotions in early childhood, whilst looking at socio-cognitive abilities as possible mechanism underlying this relation. The main findings of this study were that children with higher levels of autistic traits displayed lower levels of socio-cognitive abilities, which, in turn, were related to less expressions of guilt and embarrassment. Next to these indirect effects of autistic traits on social emotions via socio-cognitive abilities, the higher levels of autistic traits were directly related to less expressions of embarrassment, and more expressions of guilt and verbal shame-like avoidance. Autistic traits were not significantly related to displayed physical shame-like avoidance, both directly or indirectly.
Autistic Traits and Social Emotions
Children with higher levels of autistic traits showed lower levels of socio-cognitive abilities, which, in turn, was related to less expressions of guilt. This finding supports our mediation hypothesis and findings in previous studies which found autistic traits to relate to guilt through deficits in socio-cognitive abilities (Davidson et al., 2018, 2017; Muris et al., 2016). This may be because in order to experience guilt, one must have developed some socio-cognitive skills, such as perspective taking (Jankowski & Takahashi, 2014) and the awareness of transgressing social rules and norms (Davidson et al., 2017; Heerey et al., 2003a; Misailidi, 2018). However, we found that, directly, having higher levels of autistic traits was related to more guilt expressions after accounting for the indirect effect. These results indicate that for children with equally developed socio-cognitive abilities, the ones with higher levels of autistic traits displayed more guilt. Except, once the effect of socio-cognitive abilities is considered, children with higher levels of autistic traits have less developed socio-cognitive abilities, which, in turn, makes these children display less guilt. This could imply that—next to socio-cognitive abilities functioning as a negative mediator—
there also is a mechanism that we did not account for in the present study that functions positively in the relation between autistic traits and guilt. Children with high levels of autistic traits, who do not have deficits in socio-cognitive abilities, show more guilt expressions. This may be because children with high levels of autistic traits who have well-developed socio-cognitive abilities, can compensate for their socio-emotional deficits (Bauminger, 2002). Thus, it may be that their socio-cognitive skills allow them to understand they made a social transgression, and understand the social rules and norms and how they are expected to behave after a transgression. Possibly, they behave according to this “learned script” rather than the wish to repair the negative consequences of their behavior. This may explain why children with higher levels of autistic traits without deficits in socio-cognitive abilities showed more expressions of guilt. However, children with higher levels of autistic traits that do have impaired socio-cognitive abilities, may not understand they made a transgression, or may not have learned how to react to this social transgression, resulting in less guilt expressions.
Children with higher levels of autistic traits, who had lower levels of socio-cognitive abilities, displayed less embarrassment. This could suggest that the impairments in
understanding and predicting others’ mental states causes children with more autistic traits to be less aware of violating a rule, which in turn causes less embarrassment (Heerey et al., 2003a). Higher levels of autistic traits were also directly related to less displayed
embarrassment. These findings are supportive of previous studies that showed significantly less expressions of embarrassment in children with ASD (Hobson et al., 2006; Reddy et al., 2010). However, as the expression of embarrassment is indicative of appeasement and serves as a non-verbal acknowledgment of shared social standards, it is an important emotion to show for social functioning (Feinberg et al., 2012; Miller, 2007; Tangney, 1999). The direct relation between autistic traits and expressions of embarrassment implies that there is another mechanism next to low socio-cognitive abilities in play for explaining less expressions of
embarrassment in children with higher levels of autistic traits. For example, it has been stated that individuals with ASD, next to difficulties in cognitive perspective taking, have
difficulties in the affective component of empathy, which in turn may be a mechanism causing a lack of embarrassment (Baron-Cohen, 1988; Hillier & Allinson, 2002).
Consistent with previous studies, children with higher levels of autistic traits showed more verbal shame-like avoidance. However, socio-cognitive abilities could not explain this association. Alternatively, it is suggested that children with higher levels of autistic traits have elevated levels of social anxiety, including elevated fear of negative evaluations, which in turn could explain more shame-like behaviors such as avoidance (Davidson et al., 2017). Contrary to our expectations, levels of autistic traits were not associated to levels of physical shame-like avoidance. It is suggested that high levels of autistic traits in children may result in relatively unresponsive behavior to human interactions (Drash & Tudor, 2004). Possibly, this explains the different results for verbal like avoidance as opposed to physical shame-like avoidance in the present study. Basically, being verbally unresponsive—i.e., not talking after breaking Teddy—during the broken toy task, would lead to higher levels of verbal shame-like avoidance. Whereas being physically unresponsive—i.e., not doing anything including not averting or hiding after breaking Teddy—would not lead to higher levels of physical avoidance, as this required physically moving or turning away from the
experimenter. Therefore, a lack of response during interactions in children with higher levels of autistic traits could be what explains our findings for verbal and physical shame-like avoidance.
In summary, children with higher levels of autistic traits showed dysregulations in their socio-emotional expressions. Children with higher levels of autistic traits and lower socio-cognitive abilities showed less reparative and appeasing behavior after a transgression and more avoidance, at least verbally. Due to the deficits in socio-cognitive abilities, children
with higher levels of autistic traits showed a less adaptive and prosocial response when breaking a social rule. However, possibly due to learning compensational strategies, children with more autistic traits with equal socio-cognitive abilities as their typically developing peers, showed more adaptive, repairing behavior than these peers. Thus overall, having high levels of autistic traits was mostly, but not solely, related to showing less prosocial behavior and more avoidance.
Autistic traits and Socio-Cognitive Abilities
It has been stated by numerous previous studies that there is a relation between ASD and impaired socio-cognitive abilities (Baron-Cohen et al., 1985; Bauminger, 2002; Buitelaar et al., 1999; Burnside et al., 2017; Harms et al., 2010; Heerey et al., 2003a; Mazza et al., 2017; Schaller & Rauh, 2017). Consistently, in the present study, children with higher levels of autistic traits displayed lower levels of socio-cognitive abilities. This suggests that children with higher levels of autistic traits have more difficulties with the understanding and
predicting of others’ mental states. Possibly, the reduced social input because of impaired social skills and a lack of orientation to faces causes children with higher levels of autistic traits to develop lower levels of socio-cognitive abilities (Burnside et al., 2017; Harms et al., 2010; Leppänen & Nelson, 2006). The present findings contribute to the existing knowledge on autism and socio-cognitive abilities by suggesting that socio-cognitive abilities are impaired already in early childhood and also in a non-clinical sample. Additionally, the present study measured a broader spectrum of socio-cognitive abilities, including pretense, emotion recognition, emotion understanding, first order belief, and false belief.
Autistic traits and its relations to Child Age and Verbal Ability
There were some additional interesting findings with regard to child age and normed verbal ability. An important contribution of the present study is that it is first to show that associations between autistic traits, socio-cognitive abilities, and social emotions in early
childhood hold when controlling for the effects of age and verbal ability. Socio-cognitive abilities were strongly predicted by children’s age. This was expected, since socio-cognitive abilities undergo important developmental changes between the ages of 2 and 5 years. During this period, socio-cognitive abilities develop at a high pace to manifest into a ToM (Mazza et al., 2017; Poulin-Dubois & Yott, 2018; Sirois & Jackson, 2007). Additionally, guilt and verbal shame-like avoidance were predicted by child age. Older children displayed more guilt and verbal shame-like avoidance, which is consistent with the developmental characteristics of social emotions. The expression and experience of social emotions increases as our interpersonal awareness and sense of responsibility further develops (Davidson et al., 2018; Lewis, 1993, 2007; Muris & Meesters, 2014; Zinck, 2008). Moreover, guilt was positively predicted by normed verbal ability, which is consistent with the behavioral characteristics of guilt. The behaviors indicative of guilt are partly verbal expressions, so children with more verbal ability are assumed to display more guilt. However, embarrassment was negatively predicted by normed verbal ability, which could be due to the non-verbal behaviors that are indicative of embarrassment. Guilt and embarrassment are both considered prosocial
emotions, as they are indicative of appeasement and mend negative consequences in difficult social situations (Tangney, 1999). Therefore, it may be that children with less verbal ability display more embarrassment for appeasement purposes, since this requires no verbal ability, as opposed to guilt. Finally, the correlations indicated a relation between verbal shame-like avoidance and verbal ability, yet, in our final model, this relation was not found. This could be explained by the effect of autistic traits; children with more verbal ability had lower levels of autistic traits, and children with lower levels of autistic traits showed less verbal shame-like avoidance, meaning that children with less verbal ability did not necessarily show more verbal shame-like avoidance, but children with higher levels of autistic traits did.
