Mechanisms of Empathic Behavior in Children with Callous-Unemotional Traits: Eye Gaze and Emotion Recognition

Lauren A. Delk

Thesis submitted to the faculty of the Virginia Polytechnic Institute and State University in partial fulfillment of the requirements for the degree of

Master of Science In Psychology

Bradley A. White, Chair Jungmeen Kim-Spoon Susan W. White

November 16, 2016 Blacksburg, VA

Keywords: emotion recognition, prosocial behavior, callous-unemotional traits, empathy

Mechanisms of Empathic Behavior in Children with Callous-Unemotional Traits: Eye Gaze and Emotion Recognition

Lauren A. Delk

ABSTRACT

The presence of callous-unemotional (CU) traits (e.g., shallow affect, lack of empathy) in children predicts reduced prosocial behavior. Similarly, CU traits relate to emotion recognition deficits, which may be related to deficits in visual attention to the eye region of others. Notably, recognition of others’ distress necessarily precedes sympathy, and sympathy is a key predictor in prosocial outcomes. Thus, visual attention and emotion recognition may mediate the relationship between

CU traits and deficient prosocial behavior. Elucidating these connections furthers the development of treatment protocols for children with behavioral problems and CU traits. This study seeks to:

(1) extend this research to younger children, including girls; (2) measure eye gaze using infrared eye-tracking technology; and (3) test the hypothesis that CU traits are linked to prosocial behavior deficits via reduced eye gaze, which in turn leads to deficits in fear recognition. Children (n = 81, ages 6-9) completed a computerized, eye-tracked emotion recognition task and a standardized prosocial behavior task while parents reported on the children’s CU traits. Results partially supported hypotheses, in that CU traits predicted less time focusing on the eye region for fear expressions, and certain dimensions of eye gaze predicted accuracy in recognizing some emotions.

However, the full model was not supported for fear or distress expressions. Conversely, there was some evidence that the link between CU traits and deficient prosocial behavior is mediated by reduced recognition for low intensity happy expressions, but only in girls. Theoretical and practical implications for these findings are considered.

Mechanisms of Empathic Behavior in Children with Callous-Unemotional Traits: Eye Gaze and Emotion Recognition

Lauren A. Delk

GENERAL AUDIENCE ABSTRACT

Callous-unemotional (CU) traits are defined as experiencing limited emotion and empathy for others. Children with CU traits are less likely to show prosocial behavior, such as sharing or helping others. Similarly, children with CU traits also have more difficulty recognizing emotions compared to peers. This may be related to less attention to the eye region of other’s faces, which conveys emotional information. Notably, accurate recognition of others’ distress is necessary for children to feel concern for others and want to engage in prosocial behavior. Elucidating these connections furthers the development of interventions for children with and CU traits, which often related to behavior problems. This study seeks to: (1) extend this research to younger children, including girls; (2) measure eye gaze using eye-tracking technology; and (3) test the hypothesis that CU traits predict prosocial behavior deficits due to reduced eye gaze and subsequent deficits in fear and distress recognition. While this hypothesis was not fully supported, results did indicate that CU traits predicted less time focusing on the eye region for fear expressions, and certain forms of eye gaze predicted better emotion recognition accuracy for some emotions. Instead, results indicated that eye gaze and recognition of subtle happy expressions played a role linking CU traits and prosocial behavior, but only in girls. Results suggest that CU traits relate to less attention to the eye region and poorer emotion recognition across emotions, and that these mechanisms may operate differently in boys and girls.

Acknowledgements I would like to thank the children and their caregivers who took the time to participate in the study. I am also very grateful to Dr. Robin Panneton for her expert advice and her generous use of laboratory space and materials, as well as Drs. Thomas Ollendick and Alison Heck for their essential consultation in creating the protocol for this study. Additional thanks to Stephanie

Roldan and her assistance with the Matlab code for analyzing the eye tracking data. This thesis was supported with funding from the Virginia Tech Institute for Society, Culture and

Environment (ISCE) and the Center for Peace Studies and Violence Prevention (CPSVP) grants

(B.White, PI), and we thank them for supporting this study.

I am very grateful to Dr. Bradley White, my thesis chair and advisor, who has provided invaluable assistance with this project in addition to continued support throughout my time at Virginia Tech.

I would also like to thank Drs. Susan White and Jungmeen Kim-Spoon for serving on my thesis committee and offering helpful suggestions to strengthen the project. Finally, I would like to thank my husband, Kaleb, and my family for their encouragement and support.

iv

Table of Contents

Introduction ...... 1 Methods...... 12 Results ...... 20 Discussion ...... 26 Conclusions ...... 35 References ...... 37

v

List of Tables Table 1: Trial Types for Prosocial Behavior Task ...... 55 Table 2: Descriptive Statistics for Primary Variables and Accuracy ...... 55 Table 3: Descriptive Statistics of Initial Fixation ...... 56 Table 4: Descriptive Statistics of Gaze Duration ...... 57 Table 5: Paired Samples T tests for Emotion Recognition ...... 58 Table 6: Paired Samples T tests for Prosocial Behavior Task ...... 58 Table 7: Bivariate Correlations ...... 59 Table 8: Correlations with Accuracy ...... 60 Table 9: Correlations with Initial Fixation...... 61 Table 10: Correlations with Duration ...... 62 Table 11: Steiger Z tests for Significant ICU and Eye Gaze Correlations ...... 63

vi

List of Figures Figure 1: Hypothesis 1 ...... 64 Figure 2: Hypothesis 2 ...... 64 Figure 3: Hypothesis 3 ...... 64 Figure 4: Hypothesis 4 ...... 65 Figure 5: Sample AOI Placement on NimStim image (Tottenham et al., 2009) ...... 65 Figure 6: Moderated Mediation of CU traits on Recognition of Low Intensity Happy Expressions ...... 65 Figure 7: Interaction between Gender and Initial Fixation on the Eyes Predicting Accuracy of Recognition for Happy Faces at 50% intensity ...... 66 Figure 8: Moderated Mediation of Eye Gaze on Prosocial Behavior ...... 66 Figure 9: Serial Mediation Model of CU Traits Predicting Prosocial Behavior Through Initial Fixation on the Eyes of Happy Faces at 50% ...... 67

vii

Mechanisms of Empathic Behavior in Children with Callous-Unemotional Traits:

Eye Gaze and Emotion Recognition

Introduction

For a society to function successfully, antisocial behavior must be limited and prosocial tendencies (e.g., compassion, cooperation) promoted. In every culture, this struggle against disruptive behavior and toward cooperation is evidenced (Lykken, 1995), a struggle influenced not only by broader sociocultural influences, but also by differences in individual patterns of cognition, emotion, and behavior, which can manifest and change in certain ways over the course of development. Certain individual traits appear to play a particularly important role in both antisocial and prosocial tendencies, defined as actions chosen to benefit others (Eisenberg,

Fabes, & Spinrad, 2006).

Historically, Hervey Cleckley was among the first to detail in his book, The Mask of

Sanity (Cleckley, 1976), vignettes of psychiatric inpatients who seemed to “know the words but not the music” regarding healthy interpersonal feelings and interactions. These patients superficially seemed psychologically healthy, but further exploration revealed that they did not experience guilt or empathy and had remarkable histories of irresponsible, reckless behavior.

Cleckley labeled this personality style as “psychopathic.” These observations led Robert Hare to create, and later revise, the Psychopathy Checklist (PCL-R; Hare et al., 1990) which identified two distinct facets of psychopathy, interpersonal-affective deficits (e.g., superficial charm, manipulativeness, lack of remorse) and antisocial deviant lifestyle (e.g., impulsivity, irresponsibility, delinquency) which have been further differentiated and alternately conceptualized in subsequent models (e.g., Cooke & Michie, 2001).

Despite the obvious importance of psychopathy in both antisocial and prosocial tendencies, the processes by which psychopathy facets and precursors (e.g., callous-unemotional traits) are connected to the development of social behavior, particularly prosocial tendencies, is poorly understood. However, emerging evidence suggests that visual attention and emotion recognition processes may play key roles (e.g., Bowen, Morgan, Moore, & van Goozen, 2014;

Dadds, Jambrak, Pasalich, Hawes, & Brennan, 2011; Dawel, O’Kearney, McKone, & Palermo,

2012). The main objectives of the present study were therefore to address whether callous- unemotional (CU) traits predict deficits in the recognition of others’ emotional expressions (e.g., distress) by reducing visual attention (i.e. eye gaze) to the eye region of others, and whether these reductions in visual attention may lead to prosocial behavior deficits by interfering with recognition of others’ emotional expressions that otherwise could serve as social cues for prosocial responding.

Callous-Unemotional Traits

Callous-unemotional traits, which include lack of empathy, disregard for others, and lack of emotional experience, are typically considered developmental precursors to the affective core deficits observed in adult psychopathy (Frick & Ellis, 1999). CU traits are conceptualized as a constellation of behavioral patterns including uncaring (i.e., disregard for one’s performance or others’ feelings), callous (i.e., disregard for responsibilities and lack of remorse), and unemotional traits (i.e., lack of emotional expression; Byrd, Kahn, & Pardini, 2013; Frick, 2004).

They are frequently associated with trait fearlessness, lack of distress for punishment or consequences (Barry et al., 2000), and elevated sensitivity to reward (Pardini, Lochman, & Frick,

2003).

2

CU traits in youth (Ezpeleta, Osa, Granero, Penelo, & Domènech, 2013; Willoughby,

Mills-Koonce, Waschbusch, & Gottfredson, 2015), like psychopathic traits in adults (Edens,

Marcus, Lilienfeld, & Poythress, 2006), exist on a continuum in the community and appear to be important risk factors for relatively early, severe, and persistent disruptive and antisocial behavior (Frick, Ray, Thornton, & Kahn, 2014; Moffitt, 1993). By late childhood, CU traits tend to be moderately stable (Fontaine, McCrory, Boivin, Moffitt, & Viding, 2011; Frick, Kimonis,

Dandreaux, & Farell, 2003; López-Romero, Romero, & Villar, 2014; Muratori et al., 2016).

These early developing patterns increase the impetus for improving understanding of their nature, with such knowledge hopefully translating to improvement of early interventions.

Theories regarding the link between CU traits and disruptive behavior emphasize the role of deficient affective responding. Fowles and Kochanska (2000) propose that low fear response and limited affective responsivity to others increases the chances for deficient development of conscience, defined as a self-governed decision making, separate from outside influence.

Moreover, longitudinal studies indicate that moral emotions (e.g., guilt or discomfort from one’s transgressions) increase moral conduct (e.g., rule following), two primary components in the development of conscience which are predicative of self-regulation (Kochanska & Aksan, 2006).

Blair (1995) builds upon this theory when describing his Violence Inhibition Mechanism (VIM) model, which suggests that low personal distress for others reduces the likelihood that the individual will inhibit behavior that could be aggressive or otherwise antisocial towards others.

Like Fowles & Kochanska (2000), others have since proposed that CU traits relate to deficits in conscience development, which impedes empathic concern and promotes antisocial behavior

(Frick, Blair, & Castellanos, 2013).

3

In addition, there is evidence of low fear response associated with CU traits (Pardini &

Frick, 2013). Recent work supports these claims by demonstrating that CU traits are also associated with low physiological reactivity, regardless of levels of conduct problems (Fanti,

Panayiotou, Lazarou, Michael, & Georgiou, 2015). When examined in the context of limiting an individual’s response to other’s distress, fearlessness may be a path by which conduct problems develop.

Despite the term “unemotional” and fact that CU traits are typically associated with fearlessness (Fanti et al., 2015; Frick, 2012, Marsh et al., 2011), children with CU traits sometimes exhibit elevated negative affect. For instance, cluster analyses have suggested that some youth with CU traits are also high on anxiety, especially those with a history of abuse

(Kahn et al., 2013; Salihovic, Kerr, & Stattin, 2014). Furthermore, the callousness dimension in particular is positively associated with depression and anxiety (Hawes et al., 2014), and there is evidence that as CU traits and behavior problems increase, so do anxiety and depression, accompanied by decreases in self-esteem (Eisenbarth, Demetriou, Kyranides, & Fanti, 2016).

While internalizing symptoms may also accompany CU traits, internalizing symptoms are not frequently considered or controlled in studies of social information processing or behavioral problems in youth with CU traits. In such instances, the relative impact of CU traits and internalizing can be difficult to separate.

Prosocial Behavior: Nature and Correlates

While much research on CU traits emphasizes their role in disruptive behavior, they are also clearly associated with prosocial behavior deficits (Blair, 2006), as reflected in their label in the new Diagnostic and Statistical Manual of Mental Disorders (DSM-5; American Psychiatric

Association, 2013) Conduct Disorder specifier, “With limited prosocial emotions,” (Kimonis et

4

al., 2015). Reduced prosocial behavior is related to CU traits, regardless of whether or not conduct problems are statistically controlled for (Herpers et al., 2014). This finding suggests that prosocial behavior deficits cannot simply be attributed to disruptive behavior problems and must be better understood in their own right, in order to identify ways to foster prosocial behavior in youth with CU traits.

Since prosocial, like other complex behaviors, can be defined and operationalized in different ways, prior studies focus on specific forms or functions of prosocial behavior, such as prosocial reasoning (e.g., responses to transgressive behavior) or moral maturity (e.g., responses to moral statements (Herpers, Scheepers, Bons, Buitelaar, & Rommelse, 2014). Sharing resources is a specific form of prosocial behavior that develops over the course of early childhood (Fehr, Bernhard, & Rockenbach, 2008) and is itself nuanced. Economists distinguish two conditions of unequal sharing: disadvantageous inequality (DI), in which the other person in a pair receives more resources than the subject, and advantageous inequality (AI), in which the subject receives more resources than the other person (Fehr & Schmidt, 1999). In each condition

(AI or DI), sharing may come at no cost to oneself (e.g., giving away an extra ticket to a show that one would otherwise not use) or with substantial self-sacrifice or risk (e.g., entering a burning building to save a stranger). In “no-cost” type scenarios, the subject experiences no personal cost for choosing to distribute resources equally between oneself and the other person, whether the alternative choice to equal sharing is AI (getting the same amount as one would via equal sharing, but giving the other person less than oneself) or DI (getting the same amount as one would via equal sharing, but giving the other person more than oneself). In “cost” type scenarios, the participant must forfeit resources in order to achieve equal sharing between oneself and the other person, whether the alternative choice to equal sharing is AI (getting more than one

5

would have via equal sharing by giving the other person less than oneself) or DI (getting more than one would have via equal sharing by giving the other person more than oneself; Williams,

O’Driscoll, & Moore, 2014).

Thus the different conditions are separated as follows: AI no-cost scenarios compare sharing equally vs refusing the other resources so the self would benefit, AI cost scenarios compare sharing equally or benefiting notably more than the other by giving the other nothing,

DI no-cost scenarios compare sharing equally or giving the other more resources, and DI cost scenarios compare sharing equally or giving the other more resources than oneself, but also gaining more (e.g., win-win). Notably, in cases of DI, choosing equal sharing may result in withholding resources the partner would otherwise receive, which is not the most prosocial decision, even allowing for withholding resources to one’s own detriment, potentially driven by envious motives (Williams et al., 2014). Therefore, while sharing tasks typically assess perceptions of resource equality, they have also been used to measure prosocial decision making

(e.g., generosity; Shaw, Choshen-Hillel, & Caruso, 2016) in children from early through middle childhood.

Prior studies suggest that children as young as three years of age tend to find DI aversive and (across cost and no-cost scenarios) thus typically choose equal sharing in these scenarios

(LoBue, Nishida, Chiong, DeLoache, & Haidt, 2011), even giving up resources for themselves as well as for the partner in order to achieve equality (DI cost scenarios; Sheskin, Bloom, & Wynn,

2014). However, with age, children are more likely to make unequal sharing DI choices that benefit the partner (Shaw et al., 2016). To a lesser degree, children also show increased aversion to AI as they mature, such that by age 7 or 8 they increasingly share in order to avoid accruing more resources than their partner, a pattern observed slightly more often in cost trials than no-

6

cost trials (Fehr et al., 2008; Fehr & Schmidt, 1999; Sheskin et al., 2014; Williams & Moore,

2016). Notably, the context (e.g., familiarity with the sharing partner) can also influence sharing decisions and thus is typically controlled in lab-based behavioral economic paradigms, such as by enlisting an unfamiliar confederate as the partner (Williams & Moore, 2016).

Empathic Concern and Emotion Recognition

While prosocial behavior can have various motives (White, 2014), typical antecedents for prosocial behavior, particularly those motivated by genuine concern and a desire to be supportive or helpful, are: (1) the experience of empathy, defined as the ability to both comprehend the emotional state of another person (sometimes labeled perspective-taking, or cognitive empathy) and to emotionally share in their experience (sometimes termed affective empathy; Feshbach,

1978) and (2) sympathy, an other-oriented reaction to another person’s emotional state that includes feelings of concern and wanting the person to feel better (Eisenberg & Eggum, 2009).

Empathy and sympathy are generally positively related to prosocial behavior outcomes

(Eisenberg, Eggum, & Di Giunta, 2010; Malti, Gummerum, Keller, & Buchmann, 2009; Malti &

Krettenauer, 2013), which increases the importance of understanding mechanisms associated with these experiences.

The capacity to recognize others’ distress is a necessary initial step for developing sympathy for others (Eisenberg, Huerta, Edwards, 2012). Thus, the extent to which children understand other’s emotional expressions may impact prosocial behavior. Accurate emotion recognition is predictive of increased prosocial behavior such as prosocial roles (i.e. defender, mediator, and consoler by teacher report), social functioning in school, and social abilities in preschoolers (Belacchi, & Farina, 2012). Emotion recognition is also related to lower self-reports of negative social experiences in a sample of kindergarten and first graders (Miller et al., 2005).

7

Notably, accurate emotion recognition is predictive of positive social adjustment, particularly for girls (Leppänen & Hietanen, 2001). Therefore, emotion recognition is an important predictor to later prosocial behavior in a variety of contexts, including prosocial roles and social adjustment.

There is also some evidence that visual attention to others’ faces influences emotion recognition. The eyes are an especially salient region of the face for the conveyance of cues signalizing fear (Adolphs et al., 2005). Prior studies suggest that the amygdala controls visual orienting to the eye region of faces (Han, Alders, Greening, Neufeld, & Mitchell, 2012). Damage to the amygdala relates to decreased attending to the eye region and poorer emotion recognition accuracy (Spezio, Huang, Castelli, & Adolphs, 2007), suggesting that attention to different regions of others’ faces is important for emotion recognition.

Deficits in Empathic Concern: CU Traits and Emotion Recognition

Children with CU traits do not respond empathically to other’s distress, a deficit which may interfere with prosocial responding. Limited empathic concern is associated with risk for development of conduct problems (e.g, Aspan, Vida, Gadoros, & Halasz, 2013; Leist & Dadds,

2009; Sharp, 2008). Foundational studies in emotional responding in adult psychopathy demonstrated a blunted startle response and reduced autonomic reactivity to emotionally valenced images in psychopaths, compared to healthy individuals (e.g., Hare, Williamson, &

Harpur, 1988; Patrick, Bradley, & Lang, 1993).

Children with elevated CU traits demonstrate deficits identifying fearful emotional expressions in others (Blair, Colledge, Murray & Mitchell, 2001; Herpers et al., 2014), potentially due to deficits in amygdala functioning (Lozier, Cardinale, VanMeter, & Marsh,

2014; Marsh et al., 2008). Similar deficits exist in recognizing fearful body language (Muñoz,

2009). There is additional evidence that CU traits predict deficits identifying expression of

8

distress more broadly (i.e. fear, sadness; Dawel et al., 2012; Stevens, Charman, & Blair, 2001).

However, studies examining the extent to which emotion recognition deficits generalize in youth with CU traits have not to our knowledge been extended yet to younger children in early to middle childhood.

Similar to individuals with amygdala damage (Adolphs et al., 2005), individuals with CU traits demonstrate idiosyncrasies in their eye gaze which may result in deficits identifying fear or distress (Dadds et al., 2006; Marsh et al., 2008; Viding et al., 2012). Interestingly, when explicitly instructed to “look at the eyes” of static emotional faces viewed on a screen, emotion recognition improved for pre-adolescent and adolescent boys (8-17 years) with elevated CU traits toward normal levels, whereas instructions to “look at the mouth” predicted poorer fear recognition (Dadds et al., 2006). Thus, preliminary evidence suggested that CU traits in children are related to deficit fear recognition, potentially mediated through reduced attention to the eye region of others’ faces (Dadds et al., 2006). In a follow-up study, Dadds and colleagues directly examined the relationship between CU traits and eye contact directly using eye tracking (Dadds,

Masry, Wimalaweera, & Gustella, 2008). Results demonstrated that elevated CU traits predict a lower probability of initial focusing on the eye region, less time overall focusing on the eye region, and less shifting of attention back to the eye region, all predicting poorer recognition of fear expression.

Gaps in the Literature The majority of prior studies of emotion information processing deficits in youth with

CU traits have focused mostly on boys (e.g., Blair et al., 2001; Dadds et al., 2006/2008), resulting in a dearth of evidence regarding the relationships between CU traits and emotion recognition in girls. However, differences in levels of empathy and prosocial behavior are evident across genders (Hicks et al., 2012). Specifically, girls on average tend to rate negatively

9

affective stimuli as more unpleasant than boys (Sharp, van Goozen, & Goodyer, 2006). In addition, girls tend to show elevated psychophysiological responses to aggression compared to boys (Murray-Close & Crick, 2007), suggesting that elevated physiological reactions in response to other’s negative emotions may be typical of girls, and thus could influence their behavior, however, more research is necessary to clarify these relationships in females.

In addition, prior studies on relationships between CU traits and their socioemotional correlates have also tended to focus on pre-adolescent and adolescent samples (e.g. Aspan et al.,

2013; Dadds et al, 2008; Leist & Dadds, 2009; Muñoz, 2009; Sharp, 2008). Furthermore, relationships between the observed deficits in eye gaze, emotion recognition, and behavioral outcomes is underexplored. In particular, to the best of our knowledge, no published studies have examined the potential roles of eye gaze and emotion recognition in associations between CU traits and prosocial behavior deficits. Furthermore, much of the literature on behavioral problems in youth with elevated CU traits rests upon caregiver and teacher reports of behavioral problems, which may be confounded with reporting on CU traits in themselves via shared method variance.

Based on this body of literature and its limitations, the current study focused instead on a younger sample (ages 6-9 years), which allows for a downward extension of the previously observed relationships, at a time when CU traits are stabilizing but potentially still malleable

(e.g., Fontaine et al., 2011; López-Romero et al., 2014). Notably, around age six or seven, children are able to differentiate between themselves and others, show increasing perspective taking skills, and develop better understanding of various forms of reciprocity (Harter, 1983).

Moreover, children ages 5-7 and 8-9 did not have significantly different abilities in a study examining developmental changes in children’s abilities to identify mixed emotions (Zajdel,

Bloom, Fireman, & Larsen, 2013), suggesting emotion recognition skills are relatively stable

10

across the 6-9-year age range, which is consistent with by Rosenqvist and colleagues (2014) finding that the impact of age on emotion recognition begins to decelerate around 5-6 years of age. Thus, while we were able to consider age impacts if preliminary analyses suggested it was necessary, selecting this range also helped constrain the potential impact of age on emotion recognition abilities.

Current Study

The present study had several goals. First, we sought to establish if a relationship exists between CU traits and emotion recognition that predicts prosocial behavior. Next, we attempted to replicate the prior findings that CU traits predict emotion recognition deficits in a younger sample of children using a new emotion recognition paradigm. We also aimed to identify the role of initial eye gaze fixation in emotion recognition and prosocial behavior. Finally, we intended to establish a serial mediation relationship between CU traits, eye gaze patterns, emotion recognition and prosocial behavior, which has previously been shown to be negatively related to

CU traits (e.g., Herpers et al., 2014).

In summary, the four hypotheses of the current study are as follows:

 Hypothesis 1: CU traits predict less prosocial behavior due to poor emotion recognition

(Figure 1).

 Hypothesis 2: CU traits also predict failure to sufficiently visually attend (i.e. “eye gaze”) to

emotional information conveyed by others’ eye region, which in turn results in deficient

emotion recognition for distress (Figure 2).

 Hypothesis 3: Failure to attend (reduced gaze) to others’ eyes leads to reduced prosocial

behavior due to limited recognition of distress-related emotions (i.e. fear and sadness), which

mediates their relationship (Figure 3).

11

 Hypothesis 4: A serial mediation process exists, by which CU traits predict reduced prosocial

behavior through failure to sufficiently visually attend (i.e. “eye gaze”) to emotional

information conveyed by others’ eye region, which in turn results in deficient emotion

recognition for distress (Figure 4).

Methods

Participants

Participants were recruited through advertisement at day cares, camps, and through local organizations in the rural southeastern region of the United States. The final sample ranged in age from 6 – 9 years old (n=81, 34 boys, 72% White). The average age of the sample was 7.82 years (SD = 1.15 years).

Following Fritz and MacKinnon (2007), for adequate power, the final sample included 81 participants. Similar studies indicate that the standardized regression coefficients for the relationship between CU traits (or eye gaze) and emotion recognition ranged from .21 to .52

(medium to large effect sizes; Aspan et al., 2013; Dadds et al., 2008; Leist and Dadds, 2009;

Muñoz, 2008). The regression coefficients for emotion recognition predicting prosocial behavior range from .26 to .52 (medium to large effect sizes; Barth & Bastiani, 1997; Blair & Coles, 2000;

Leppänen & Hietanen, 2001). Therefore, assuming the medium effect size for both paths a and b, based upon the medium effect size observed in prior studies, Fritz and MacKinnon’s (2007) article suggests a sample size of 78 for the percentile bootstrap test for a simple mediation, without covariates. Thus, the current study was adequately powered to detect a moderate mediation effect in the initial models, but underpowered to include covariates or detect moderate serial mediation effects.

12

Procedures

IRB approval was obtained prior to recruitment or data collection. Recruitment entailed posting flyers in community centers and day-cares in the region, posting electronic notices, and sending mailing to potential participants, identified through a departmental database of individuals who previously consented to be contacted with information for research studies.

Parents or guardians who were interested in participating contacted the investigator and completed a phone screen, including verbal consent, prior to scheduling the laboratory session.

Information on current diagnoses were obtained, and if the child required corrective lenses the parent was notified that glasses may interfere with tracking, thus disqualifying participants.

Exclusion criteria included a prior diagnosis of Autism Spectrum Disorder, Intellectual

Disability, or Bipolar Disorder, because such diagnoses have significantly impact on emotion recognition abilities. In addition, recruitment was monitored to achieve a gender balance within the sample (no more than a 60-40% difference) to increase the odds of identifying gender differences where they exist.

As an incentive and reimbursement for participating, caregivers received $20 each for participation in this study and children received a small toy from a prize box. Following informed consent and assent procedures, caregivers completed questionnaires about their children’s demographics and behavior, including the Inventory of Callous-Unemotional Traits

(ICU; Frick, 2004), while children completed the emotion recognition and prosocial behavior tasks.

Similar to Dadds and colleagues (2008), children’s visual attention was tracked using a

Tobii T60 infrared eye tracker with 17-inch display, maximum resolution of 1,280 x 1,024 pixels, and a sampling rate of 60 Hz. Each participating child was seated at a fixed distance of

13

approximately 60 cm in front of the eye-tracker monitor and a five-point calibration was completed, followed immediately by the emotion recognition task. Afterward, the child was escorted to a neighboring room for the prosocial behavior task.

Measures

Inventory of Callous Unemotional Traits – Parent Version (ICU; Frick, 2004). The

ICU is a 24 item questionnaire with a 4-point Likert scale. The ICU was designed to improve upon several psychometric aspects of the Antisocial Process Screening Device (APSD; Frick &

Hare, 2002) and is routinely used in assessment of CU traits in children (Frick, 2004). Although few studies have used the parent-report version of the ICU, its use is on the rise (Benesch, Görtz-

Dorten, Breuer, & Döpfner, 2014). The ICU has good internal consistency across 18 months at six-month intervals (Cronbach’s αs ranging .86 to.87) for children ages 6-12 years (Fanti &

Centifanti, 2014; Kimonis, Fanti, & Singh, 2014). In addition, in a recent meta-analysis of studies examining CU traits (Waller, Gardner, & Hyde, 2013), most (n=23 of 30, 76.7%) used parent report to measure CU traits, most often through items on the ASPD. Moreover, of those studies that used parent report to measure CU traits, 18 overlapped with the targeted age range of this study.

Research on the internal structure of the ICU has produced mixed results. While many studies support a three-factor bifactor model (Essau, Sasagawa, & Frick, 2006; Ezpeleta et al.,

2013; Frick & Ray, 2014), reflected by one overarching general factor and three underlying specific factors (uncaring, callous, and unemotional; e.g.,), more recent studies support various

2-factor models based on shortened versions of the ICU (Hawes et al., 2014; Houghton, Hunter,

& Crow, 2013; Waller et al., 2015; Willoughby et al., 2015). Still others highlight concerns about method variance on the ICU that may affect apparent factor structure, leading to

14

recommendations to focus on the ICU total score (Paiva-Salisbury, Gill, & Stickle, 2016; Ray,

Frick, Thornton, Steinberg, & Cauffman, 2016). Thus, the present study completed analyses with the total score for the ICU. In the current study, Cronbach’s α = .86 for the ICU total score.

Child Behavior Checklist (CBCL; Achenbach & Rescorla, 2001). The CBCL/6-18 is a parent-report instrument commonly used to assess psychopathology in youth. It consists of 113 questions, scored on a three-point Likert scale (0=absent, 1= occurs sometimes, 2=occurs often).

The Anxiety Problems scale was used to assess whether internalizing symptoms may confound relationships between primary variables of interest (CU traits, emotion recognition, eye gaze, and prosocial behavior) in light of prior research showing that anxiety can co-occur with CU traits and thus may impact behavior (Hawes et al., 2014; Kahn et al., 2013; Salihovic et al., 2014).

Sensitivity to Punishment and Sensitivity to Reward Questionnaire for Children

(SPSRQ-C; Colder & O'Connor, 2004). This questionnaire measures reinforcement sensitivity in children through parent report and contains 33 items. It comprises a Punishment Sensitivity scale, and three Reward Sensitivity scales: Reward Responsivity, Impulsivity/Fun-Seeking, and

Drive. Each item is scored on a 5-point Likert scale (1=strongly disagree, 5=strongly agree). It has demonstrated good internal reliability for all four scales in children ages 6-13 (Luman, van

Meel, Oosterlaan, & Geurts, 2012). More recent analyses identified a Fear/Shyness scale (Colder et al., 2011), which was used to test fearfulness as a potential confound, in light of prior evidence that low fear response is associated with both CU traits (Fanti et al., 2015; Pardini & Frick,

2013), as well as development of conscience which impact prosocial behavior outcomes (Fowles

& Kochanska, 2000). Thus, a child’s level of fear may act as a confound when examining these variables. In the current study, α = .87 for the SPSRQ-C Fear/Shyness scale score.

15

Emotion recognition task. To assess emotion recognition, we created a new computerized emotion recognition paradigm based on the UNSW Facial Emotion Recognition

Task (Dadds, Hawes, & Merz, 2004). However, because stimuli for that task were not standardized, which could confound emotion recognition, we elected to instead use previously validated stimuli from the NimStim set (Tottenham et al., 2009) that had been rated at 71% or higher agreement that the image was representative of the target emotion. Three adult male faces and adult three female faces were selected that were rated highly on each emotion (using happy, fear, anger, sad, and neutral) expression, counterbalancing for mouth position (open vs closed) across genders and emotions. There were 54 trials in total, with nine trials per model (four at

50%, four at 100% intensity, and one neutral face). Only Caucasian faces were used in this initial investigation to reduce potential confound of implicit racial bias.

Recently, multiple researchers have created their own stimuli at various intensity levels in order to reduce the likelihood of ceiling effects (Gao, Maurer, & Nishimura, 2010; Gillespie,

Rotshtein, Satherly, Beech, & Mitchell, 2015; Tell, Davidson, & Camras, 2014). For this study, we followed the protocol described by Calder, Young, Perrett, Etcoff, and Rowland (1996) and used the Morpheus Photo Morpher (as used by Tell et al., 2014) to ‘morph’ the prototypical images of a neutral expression with each of the other emotional expressions, selecting the frame associated with 50% of the full emotion and the neutral emotion combined. The final task included 50% intensity trials followed by 100% intensity trials. We further adapted it for use with younger participants by requiring a verbal numeric response to a visual response set

(“Which number matches how they are feeling?”), rather than having participants name the expressions (Dadds et al., 2006). This modification was implemented to permit measurement of emotion recognition while controlling for individual differences in ability to accurately label

16

emotions. A single female face rated most gender neutral by three independent judges was used to display all emotions in the response key. We counterbalanced response key display order (emotions labeled 1and 2 on top and 4,5,6, directly below them) across trials of the same emotion within each intensity block (50% and 100%), in order to avoid confounding emotion type with response location.

Immediately prior to the emotion recognition task, the participants completed a shape- matching task on the same Tobii monitor to ensure they understood the directions for the emotion recognition task. Each image was displayed for 2 seconds and the response options were displayed visually by including a range of emotional faces from a single model, paired with corresponding number choices. During the task, the participants responded verbally with the name of the number corresponding to the selection. Correct identification was coded as 1, incorrect identification was coded as 0, and accuracy scores were computed by averaging the points across trials for each emotion.

Eye tracking. Following Dadds and colleagues (2008), we examined on each Emotion

Recognition Task trial the location of initial fixation and duration of fixation to each facial stimulus as indices of eye gaze patterns. In the literature it is most common to define a fixation as eye gaze confined for at least 100 ms to a given pixel radius or within an Area of Interest

(AOI). This standard has been used in participants from 4 months to 24 years (Amso, Haas, &

Markant, 2014; Best, Yim, & Sloutsky, 2013; in 9-17 year olds; Guyer et al., 2008; Hunnius, de

Wit, Vrins, & von Hofsten, 2011). Individual trials missing more than 50% of tracking data were coded as missing data for that participant (7.36% of all trials). Three cases were removed from the data set prior to analyses due to significant data loss (i.e., less than 2 complete trials for an emotion). Following methodology from similar studies (Leitzke & Pollak, 2016) AOI’s included

17

the eyes, mouth, or the remainder of the face region (Figure 5) with the eye region defined by an ellipse from the upper nose to the brow and from temple to temple and the mouth region defined as the area immediately surrounding the mouth and lips extending up to just under the tip of the nose. Initial fixation and duration scores were derived by averaging these scores across trials for each relevant emotion. Data reduction code was developed in MATLAB.

Prosocial behavior task. In order to assess prosocial behavior through a sharing paradigm, we followed a protocol for emotion induction set forth by Williams and colleagues

(2014) which was previously validated in children as young as three. Using the identical stimulus video from this study, the children were exposed to an empathy-induction scenario. Specifically, the video detailed the story of another child (“Jenny”) who first plays with her dog, but then she describes in a sad tone that he ran away. She is visibly upset as she makes posters and looks for her dog. The child participant was then asked how Jenny felt and to reflect on how he or she felt during the video, before completing the resource-allocation (sticker-sharing) task. After completion of the resource-allocation task each child was reassured that Jenny’s dog returned home.

During the resource-allocation task, which was previously validated for this age range

(Fehr & Schmidt, 1999; LoBue, Nishida, Chiong, DeLoache, & Haidt, 2011; Malti et al., 2009;

Sheskin, Bloom, & Wynn, 2014), the children were presented with several dichotomous choices in which they decided how many stickers to keep for themselves and how many to ostensibly mail to Jenny. These trials were divided into two groups with two subtypes each, AI cost and no cost as well as DI cost and no cost. All trials were scored such that the more prosocial decision earned one point and less prosocial decisions earned no points. In AI no cost trials the child can choose to give Jenny a sticker to make it equal (vs one sticker for themselves and none for

18

Jenny). In the AI cost trials, the child had to choose to give up an extra sticker for themselves in order to achieve equality (by giving one sticker to Jenny). In the DI no cost conditions, the child could choose to give more stickers to Jenny (although they received no more stickers themselves, thereby violating equality). Whereas in the DI cost condition they could give Jenny two more stickers and receive an extra sticker themselves (again, violating equality but a more prosocial choice). Each trial was paired with an equal option (one sticker each). Each child completed 4 rounds of each trial type which were counterbalanced over two blocks, separated by a coloring task (Table 1). The children placed their selection for the other child into envelopes to encourage a sense of privacy and reduce the likelihood of response bias.

Data Analysis

First, univariate outliers were identified based on a criterion of 3 SDs above or below the mean. These outliers were winsorized to a value of 3 standard deviations from the mean. The data were then examined for multivariate outliers using Mahlnobis distances, although there were no outliers identified. All predictor variables demonstrated limited deviance from normality except for initial fixation on the mouth for angry (50%) faces and accuracy for happy (100%) faces (George & Mallery, 2010). Bivariate correlations were examined to consider potential confounds between variables of interest (CU traits, eye gaze, emotion recognition, and prosocial behavior) and common correlates (gender, age, race, anxiety, and fearfulness), and to examine at the bivariate level the hypothesized relationships between CU traits, emotion recognition, eye gaze patterns (fixation and duration), and prosocial behavior. Paired sample t-tests were completed for emotion recognition scores to determine if general distress and total emotion recognition scores could be computed (vs. examining each emotion separately). Similarly, paired

19

sample t-tests were computed for the prosocial behavior task scores to determine if the different trial types should be combined.

Multiple regression analyses were used to examine the relationships between scores on the ICU, the point of initial eye gaze fixation, accuracy of emotion recognition and prosocial behavior. Hayes’ PROCESS macro (2013) was used to test mediation hypotheses by bootstrapping estimates of hypothesized indirect effects. In each mediation model, “Path c” refers to the total effect of X on Y, “Path c'” refers to the direct effect of X on Y when M is included in the model, and with regard to the indirect (mediation) effect, “Path a” refers to the effect of X on M, and “Path b” to the effect of M on Y (controlling for X), where X is the independent variable, M is the moderator, and Y is the dependent variable of interest.

Results

Descriptive statistics for the ICU, demographics, the prosocial behavior task, and accuracy of the emotion recognition task are presented in Table 2. Descriptive statistics for initial fixation and gaze duration are presented in Tables 3 and 4, respectively. ICU mean total scores

(M = 16.88, SD = 8.06; Table 2) followed similar patterns to prior studies using the parent-report form of the ICU (Kimonis et al., 2014). In addition, 13 children (6 boys, 7 girls; 16% of the final sample) surpassed the clinical cutoff of 24 points, indicating parental report of significant levels of CU traits (Kimonis et al., 2014).

Mean accuracy of emotion recognition ranged from 51% (for sad expressions) to 74%

(fear expressions) for the 50% intensity stimuli (Table 2). As expected, mean accuracy for emotion recognition was higher for the 100% intensity expressions, ranging from 77% (fear/sad) to 93% (happy). T-test comparisons of means accuracy across emotion expressions for 50% intensity stimuli indicated that recognition for each emotion expression type significantly

20

differed from one another (all ps < .01) except for fear and angry expressions (p = .969; see

Table 5). For the 100% stimuli, recognition of sadness was higher than at 50% and not significantly different from fear recognition at 100% intensity. Also at 100% intensity, angry expression accuracy scores were not significantly different from recognition of happy expressions but did show a nonsignificant trend (p = .080) with recognition for happy expressions being marginally higher than for angry faces. Eye gaze patterns to the eye AOIs and mouth AOIs differed in some instances across emotions and intensities. For 50% intensity, duration of gaze to the eye region AOI and to the mouth region AOI differed across several but not all emotions. Specifically, participants fixated longer on the mouth for sad compared to fear expression (p < .001). Initial fixation to the eye or mouth AOI did not sig differ for any emotions at 50% intensity (ps > .081). For emotions at 100% intensity, duration and initial fixation on eye

AOIs did not differ based on emotion (all ps > .070). However, participants spent more time looking at the mouth during fear compared to sad expressions at 100% intensity (p = .013), and they were more likely to look at the mouth first during fear compared to sad expressions at 100% intensity (p = .020). Due to the observed significant mean differences in emotion recognition accuracy and in eye gaze patterns across the emotional expressions and intensity conditions, separate analyses were run for each emotional expression and intensity level, rather than combining scores to compute general distress (fear + sad expressions) or total emotion recognition score.

Next, a paired samples t-test was completed for the AI and DI conditions to compare scores on the trial types (cost vs. no cost) comprising these conditions. In our study, the means of the cost vs no-cost trial type in both AI and DI conditions were significantly different (Table 6).

These significant differences among trial types across conditions and their medium correlations

21

with one another were similar to those of other studies (Williams & Moore, 2016). However prior studies only use one trial type (cost or no cost) to represent each condition (AI or DI), or did not test for differences among the trial type means (LoBue et al., 2011; Sheskin et al., 2014).

Williams and Moore’s (2016) primary research question regarded preference for equality and thus they argued that the intercorrelations were high enough to justify combining the trial types, in spite of the statistically significant differences in means (r’s for AI trials ranged from .40 to

.58, r’s for DI trials ranged from .47 to .64). We instead separated the trials to permit consideration of potential distinctions in prosocial behavior among the four (AI/DI; no-cost/cost) sharing scenarios.

Bivariate correlations (Table 7) were considered next, first to determine whether potential confounds required statistical control. The CBCL anxiety problems scale significantly correlated with ICU total scores (r = .24, p = .032), but not with outcome variables of interest (i.e., emotion recognition accuracy, AI, or DI scores). Similarly, gender related to some gaze patterns, but not to any outcome variables of interest. Age significantly correlated with DI no-cost and DI cost scores (r = .36, p = .001 and r = .32, p = .004, respectively) as well as several measures of fixation and accuracy, and thus was included as a covariate in the associated analyses. The fear/shyness scale only correlated with recognition of angry faces at 50% intensity (r = -.27, p =

.014; Table 8).

In relation to our primary hypotheses, at the bivariate correlation level, ICU total scores did not correlate with any of the four prosocial behavior (sticker sharing) types (all ps > .32).

ICU total scores were negatively associated with recognition of happy faces at 100% intensity (r

= -.28, p = .012; Table 8), but did not correlate with recognition of fear or any other emotional expression types at either intensity (all ps > .05). Notably, nonsignificant trends (p < .1) included

22

the relationship between ICU total scores and recognition of 50% happy expressions (r = -.19, p

= .082), and ICU total scores and recognition of 100% sad expressions (r = -.19, p = .09). ICU total scores negatively correlated with initial fixation on the eyes only for happy and sad expressions at 50% intensity (r = -.23, p = .037, and r = -.25, p = .024, respectively, Table 9; all other ps > .09). A nonsignificant trend occurred between ICU total scores and initial fixation on the mouth for 50% angry expressions (r = .19, p = .094). In addition, ICU total scores significantly correlated with duration of gaze to the eye region for fear and sad expressions at

50% intensity (both rs = -.25, p = .027 and .024, respectively, Table 10, all other ps > .111).

Furthermore, emotion recognition for any emotional expression at either intensity did not correlate with any prosocial behavior type; (all ps > .05; Table 7).

Regression results for our mediation hypotheses were considered next. The first hypothesis was that CU traits predict poorer emotion recognition, which in turn predicts prosocial deficits. This mediation model was not supported for fear expressions. Specifically, there was no evidence of a total effect of CU traits on prosocial behaviors (Path c: all ps > .305) or on emotion recognition for fearful faces at either 50% (Path a: p = .128) or 100% intensity (p

= .279). Furthermore, emotion recognition for fearful faces at faces at either 50% or 100% intensity failed to predict any of the four sharing behavior scores (Path b: all ps > .617), and the

95% C.I.s of the bootstrapped estimate of the indirect effect contained zero, thus failing to support mediation. Results were similarly nonsignificant for the same model for sad faces (as the other component of distress-related emotional expressions), as well was for other emotional expressions considered (happy, angry).

Our second hypothesis was that deficient visual attention to others’ eyes mediates the relationship between CU traits and emotion recognition deficits. This mediation model was not

23

clearly supported for fear expressions, although some pathways in the model were supported.

Specifically, there was no evidence of a total effect of CU traits on emotion recognition for fearful faces at either 50% (Path c: all ps > .128) or 100% intensity (all ps > .279). ICU total scores did inversely predict as expected children’s duration of fixation on the eye region for fearful faces displayed at 50% intensity (Path a: b = -7.6135, p = .027), but ICU total scores did not predict duration of fixation on the eyes for fear faces at 100% intensity, or initial fixation on the eyes at either 50% or 100% (all ps > .280). Also, duration of gaze to the eye region for fearful faces inversely predicted as expected children’s emotion recognition for fear expressions displayed at 100% intensity (Path b: b = -.0003, p = .030) but not for fear at 50% intensity or for initial fixation at either 50% or 100% intensity (all ps > .077). Furthermore, the 95% C.I.s of the bootstrapped estimate of the indirect effect contained zero, thus failing to support mediation.

For sad expressions, as the other component of distress-related emotional expressions, results indicated that ICU total scores predicted both duration and initial fixation on the eyes for sad expressions at 50% intensity (Path a: b = -7.8802, p = .024, and b = -.0080, p = .024, respectively). Otherwise, results were nonsignificant for the remained of the model for sad faces, as well was for angry expressions.

Despite these null findings of the second hypothesis for fear and sadness expressions, mediation of the CU trait – emotion recognition relationship by eye gaze was supported for happy expressions at 50%. Specifically, as illustrated in Figure 6, the relationship between ICU total score and recognition of happy faces at 50% intensity (Path c: b = -.0056, p = .082) was qualified by a significant moderated mediation, with ICU total score predicting fixation to the eyes (Path a: b = -.0066, p = .0370, and the relationship between fixation to the eyes and

24

recognition of happy expressions at 50% moderated by gender (Figure 6 & 7). The indirect effect

(Path c') was only significant for females (Figure 6; b = -.0037, 95% CI [-.0078, -.0004]).

The third hypothesis was that less gaze to the eye region of others relates to poorer recognition of fear emotions, which in turn predicts prosocial deficits. This mediation model was partially supported for fear expressions. While there was no evidence of a total effect of limited gaze to the eye region predicting prosocial behavior deficits (Path c: all ps > .144), there was evidence of an effect of duration of gaze to the eye region on emotion recognition for 100% intensity fearful faces (Path a; b = .0003, p = .022) similar to Path b in hypothesis 2. However, as in hypothesis 1, emotion recognition for fearful faces at faces at either 50% or 100% intensity failed to predict any of the four sharing behavior scores (Path b: all ps > .354), and the 95% C.I.s of the bootstrapped estimate of the indirect effect contained zero, thus failing to support mediation. Results were similarly nonsignificant for the same model for sad faces (as the other component of distress-related emotional expressions), as well was for angry expressions.

As in hypothesis 2, while there was not substantial evidence for relationships with fear or sad expression, the third hypothesis was supported for happy expressions. Gender again significantly moderated the relationship between fixation and recognition in a moderated mediation of the relationship between first fixation on the eyes and prosocial behavior for DI no cost trials (Figure 8). As before, this relationship was only significant for girls (b = .7014, 95%

CI [.0277, 1.7092]).

Our final hypothesis 4 proposed that a serial mediation process exists, in that CU traits predict reduced prosocial behavior via failure to attend to the eyes (reduced eye gaze) which in turn results in deficient emotion recognition for fear. This mediation model was only partially supported for fear expressions. Specifically, there was no evidence of a total effect of CU traits

25

on prosocial behaviors (Path c: all ps > .319). However, as noted earlier, there was some evidence that CU traits predict less time focusing on the eyes for 50% fear faces (b = -7.6135, p

= .027; also Path a of hypothesis 2), as well as a relationship between duration of gaze to the eye region and emotion recognition for 100% intensity fearful faces (b = .0003, p = .03; also Path b of hypothesis 2). Regardless, emotion recognition for fearful faces at either 50% or 100% intensity failed to predict any of the four sharing behavior scores (Path b: all ps > .429), and the

95% C.I.s of the bootstrapped estimate of the indirect effect contained zero, thus failing to support mediation. Results were similarly nonsignificant for sad and angry expressions.

On the other hand, serial mediation was supported for recognition of happy faces at 50% intensity (Figure 9). The ICU total score predicted the probability of first fixating on the eyes for happy faces at 50% intensity (b = -.0066, p = .037). In turn, first fixation on the eyes predicting recognition of happy faces at 50% intensity (controlling for ICU total score, b = .2375, p = .038).

Finally, controlling for all previous variables, recognition for happy expressions predicted prosocial behavior specifically in the DI no cost trials (b = 1.2949, p = .068). The indirect effect of this serial mediation was significant (Path c': b = -.0022, 95% CI [-.0105, -.0001]). As there was evidence for gender moderation of the path from eye gaze to emotion recognition, the full model in Figure 9 is likely primarily accounted for by this association in females, although we did not test for moderation of this pathway due to limited power.

Discussion

In the current study, we proposed a series of mediation analyses, based on prior literature, that CU traits in young boys and girls are associated with prosocial behavior deficits

(specifically, sharing with a sad, same-aged peer) because youth with CU traits have difficulty

26

visually attending to the eye region of others’ faces, thus impeding recognizing distress in others.

This study posed several specific hypotheses to test this model.

With regard to the first hypothesis, we predicted that that elevated CU traits in children relate to poorer emotion recognition, compared to children with low CU traits (Dadds et al.,

2008), thereby predicting less prosocial behavior. However, this hypothesis was not supported.

First, CU traits did not predict lower levels of prosocial behavior on the resource allocation task.

This null result may have been due to social desirability, as the child participants made their sticker-sharing selections on the prosocial behavior task in front of the experimenter (Heyman,

Barner, Heumann, & Schenck, 2014). Alternatively, it may be because prosocial behaviors and the processes driving them (e.g., sympathy, moral motivation; Malti, Gummerum, Keller, &

Buchmann, 2009) have a variety of influences, not just visual attention and emotion recognition.

Alternatively, it could have been due to the limited number of children with elevated CU traits in the sample. Notably, while the current study employed regression analyses, Dadds and colleagues (2008) used an ANOVA-based approach including only youth scoring high (75th percentile) and low (25th percentile) on CU traits, which may have impacted results. In addition, studies of CU traits assess for them in various ways (Waller et al., 2013), and the approach by

Dadds and colleagues (2006, 2008) differed from our own. These differences, as well as differences among the samples (age and gender) may have resulted in divergence between our findings and those of Dadds and colleagues with regard to CU links to visual attention as well, described below.

Another potential explanation for null results is children’s aversion to inequality, (Fehr et al., 2008; Williams & Moore, 2016). This literature suggests that, around age seven, children avoid both inequality that favors them (advantageous) and inequality that favors others

27

(disadvantageous; Fehr et al., 2008). However, children’s reactions to DI may change throughout development. Williams and Moore (2016) suggested that aversion to DI may stem more from envy as children age, but also note that this relationship may be qualified by one’s relationship to the interactional partner (e.g., friend vs. stranger). Therefore, it could be that the desire for equality has more influence on sharing behavior than CU traits do in this age range and context.

Also in this sample, there were significant relationships between DI cost and no cost trials with age, indicating that older children were more likely to make prosocial sharing decisions, regardless of CU trait levels. Thus emergence of individual differences related to CU traits in prosocial responding on this resource allocation task might be more discernable in an older sample.

Our second hypothesis stated that CU traits would relate to reduced visual attention to the eye region of others’ faces, resulting in poorer emotion recognition. We anticipated this relationship to be strongest for fearful expression recognition, based on prior work by Dadds and colleagues (2006, 2008). While CU traits did not predict reduced initial fixation on the eyes for fear faces, in contrast to findings by Dadds and colleagues (2008), children with elevated CU traits were significantly less likely to first fixate on the eyes for relatively subtle (i.e. 50% intensity) happy and sad expressions. CU traits also predicted less overall time fixating on the eyes for subtle fear and sad expressions. These results appear to be in support of arguments that

CU traits are predictive of deficient attention and emotion recognition related to all emotions, not just fear (Dawel et al., 2012; Dawel et al., 2015). Notably, CU traits were not related to increased attention to the mouth region, replicating Dadds and colleagues (2008) conclusions that CU traits are predictive of reduced attention to the eyes, rather than increased attention to the mouth region per se. In addition, it is interesting that these relationships were only significant for subtler (50%

28

intensity) stimuli. This may suggest that individuals with CU traits are most vulnerable to overlooking subtle emotion cues (Blair et al., 2004; Bowen et al., 2014). However, when comparing ICU score to gaze patterns for 50% intensity stimuli with the similar correlations for full-intensity stimuli, only duration of gaze to the eyes for sad faces (at partial and full intensity) had significantly different relationships with ICU scores (Table 11.; Steiger, 1980). Thus, there is not enough evidence in the present study to support the theory that CU traits predict deficient attention for subtle emotional cues, but future research should consider and test such possibilities.

The second part of this second hypothesis was partially supported, in terms of gaze patterns predicting emotion recognition. While first fixation on the eyes for fear expressions did not predict poorer accuracy, more initial fixation on the mouth region predicted poorer recognition of fear faces for 100% intensity. While unexpected, initial fixation on the eyes for relatively subtle happy expressions predicted improved recognition for these expressions. Also consistent with hypotheses, duration of fixation on the eyes for fear faces predicted better recognition for full-intensity fear expressions. In our stimulus set, the whites of the eyes are notably less pronounced in subtle fear expressions than full-intensity fear expressions, and perhaps provide less relevant emotional cuing information to participants. This positive relationship between gaze and emotion recognition for full emotions supports the notion that widened eyes, allowing more visible sclera, are important for recognition of fear expressions

(Dadds et al., 2011).

Support for the proposed mediation model, that CU traits would predict reduced eye gaze resulting in poorer emotion recognition, was only found for subtle happy expressions.

Furthermore, the relationship between CU traits and emotion recognition via reduced initial

29

fixation to the eye region was only significant for girls viewing subtle happy expressions. Gender moderated the path between initial fixation and emotion recognition, indicating a significant relationship for girls but not boys in this sample. Notably, mean scores for emotion recognition were not significantly higher for girls than boys in this sample. It may first appear contradictory that initial fixation on the eyes improves recognition for happy expressions, since happy expressions are commonly denoted by a grinning expression. However, Frank, Ekman, and

Friesen (1993) suggest that there is important information conveyed in the orbicularis oculi action (eye muscles), such that movement around the eyes conveys additional salient emotional information accompanying a smile. Furthermore, the relatively subtle (50%) happy faces did not show a smile as clearly as in the full-intensity stimuli. Thus girls may be better at detecting tension in these ocular muscles as clues to the emotion than boys. We revisit discussion of gender differences in interpreting findings for our third hypothesis.

Our third hypothesis stated that less attention to the eye region would predict less prosocial behavior through poorer emotion recognition. While a link between gaze and emotion recognition was not evidenced for fear or sad expressions, there was a significant relationship for subtle happy faces. This relationship was qualified by a gender interaction, in that girls who first fixated on the eyes of subtle happy expressions were more likely to correctly identify the happy expression, whereas there was no relationship for boys. This suggests that females may have some advantages in emotion recognition, although such gender differences are likely small

(Lawrence, Campbell, & Skuse, 2015). The significant moderated mediation model demonstrated that attention to the eye region for subtle happy emotions statistically predicts more prosocial behavior from girls when presented with the opportunity to share with a peer.

Thus, eye gaze potentially explains why emotion recognition would predict prosocial outcomes

30

for girls (Leppänen & Hietanen, 2001). In addition, there is substantial prior work which indicates that girls report more affective empathy than boys (Michalska, Kinzler, & Decety,

2013) and that girls are more likely than boys to demonstrate prosocial behavior with same-sex peers (Feshbach, & Roe, 1968; Stuijfzand, De Wied, Kempes, Van de Graaff, Branje, & Meeus,

2016).

Notably, our selected prosocial behavior task focused upon responding to a sad rather than fearful child, which may have impacted our findings. Furthermore, while Kochanska (1993) and others have emphasized the ability to experience others’ distress as an antecedent to prosocial behavior, happy expressions also serve communicatory and reinforcement functions of increasing the probability of one’s behaviors that elicit others’ happiness (Blair, 2003; Matthews

& Wells, 1999). We thus propose that, in addition to recognizing others’ distress, recognizing others’ positive affect may also motivate prosocial behavior by allowing others’ positive affect to serve as a reinforcing social cue when one acts in a caring or supportive manner toward others.

Interestingly, exogenous administration of oxytocin may improve both recognition of others’ happiness (Marsh, Henry, Pine, & Blair, 2010) and prosocial behavior (Campbell, 2010), although these relationships may be specific to individuals with existing social-emotion deficits and particular contexts (Bartz, Zaki, Bolger, & Ochsner, 2011). There is also some evidence that recognition of both happy and sad expressions increases approach behavior (Mizokawa,

Minemoto, Komiya, & Noguchi, 2013; Seidela, Habela, Kirschnerc, Gurd, & Derntl, 2010), with approach being required for many (although not all or exclusively) prosocial responses, including sharing of resources. Therefore, if one cannot recognize the positive impact of one’s prosocial behavior on others via their displays of happiness, then one may be less likely to engage in prosocial behavior. Consistent with this conjecture, we observed in the present study

31

that in girls specifically, less attention to the subtle information from the eye muscle region appeared to potentially reduce recognition of happiness, and was associated with a lower likelihood of choosing a prosocial response, even when there is no difference in resources they themselves would receive (i.e., DI no cost scenario). For the girls with better eye contact, the relationship was reversed, and associated with higher levels of prosocial behavior.

The hypothesized serial mediation model was only supported for first fixation to the eyes of subtle happy expressions predicting less prosocial decisions during no-cost disadvantageous inequality trials. This is contrary to prior work, which indicates that children in this age range are more likely to choose the equal-sharing (i.e. less prosocial) option during no cost trials (Williams et al., 2014; Williams & Moore, 2016). However, there was a significant relationship between

CU traits and reduced initial fixations on the eyes, suggesting that in both boys and girls, CU traits predict less attention for the eye region while viewing subtle happy faces. The significant gender moderation suggests that this serial mediation is primarily accounted for by the relationship between initial fixation and emotion recognition in girls. Notably, there were not differences in mean accuracy based on gender. Therefore, initial fixation on the eyes did not improve girls’ recognition of low intensity happy faces beyond boys’ abilities, but there was a significant positive relationship. In turn, girls who did not initial fixate on the eye region were less likely to choose prosocial sharing even though it did not change the outcome for themselves.

According to theories of conscience, their actions may be driven by limited feelings of sympathy when confronted with the lost dog scenario (Eisenberg & Eggum, 2009; Eisenberg et al., 2012;

Fowles, & Kochanska, 2000).

There are several noteworthy limitations of the current study. First, it is cross-sectional.

Others have inferred (Dadds et al., 2011; Viding et al., 2012) that attention to the eyes must

32

occur before recognition, and emotions must be recognized before an empathic prosocial response can be generated, implying causal directionality. However, temporal precedence cannot be formally tested in the current study. More so, although CU traits are often treated as predisposing temperamental features, there is not definitive evidence of what point in development these traits arise or are solidified, and what processes are involved. Thus, this cross- sectional study cannot make conclusions as to the impact of CU in the manifestation of current eye gaze patterns. Alternative plausible scenarios include deficient eye gaze and emotion recognition leading to CU traits.

Another limitation of this study is the sample size, which limited power, reducing the likelihood of detecting smaller effects, including moderating or serial mediation effects, increasing the risk of type II errors. Conversely, the number of correlations tested inflates type I error rate. If Bonferroni corrections were made the following relationships would remain significant: the relationship between first fixation on the mouth and emotion recognition for fear expressions at 100% intensity (r = -.38, p = .001), as well as the relationship between age and DI no cost and cost trials (r = .36, p = .001 and r = .32, p = .004, respectively). Given that most of the standardized regression coefficients for the mediation paths in this study ranged from .15 to

.25, according to Fritz and MacKinnon (2007) we would need over 400 participants to achieve power of .80. Determining the necessary N for a serial mediation is more complicated, given that even if individual paths are highly powered, the mediated effects are often small and thus tests of them are often underpowered (Thoemmes, MacKinnon, & Reiser, 2010).

In this preliminary investigation, emotion recognition accuracy was considered for each expression without consideration to the nature of mistakes made. Future studies should examine

33

confusion matrices to compare error patterns of emotion categorization and consider potential systematic recognition errors (e.g., mistaking fearful for angry expressions).

Finally, this study did not take into account the role of different forms of prosocial behavior (Dunfield, 2014) and that such subtypes may relate differently to both emotion recognition and the context of the sharing task. Even from infancy, neural correlates relate differently to helping compared to comforting behaviors (Paulus, Kühn-Popp, Licata, Sodian, &

Meinhardt, 2013). As this study only focused on one form of prosocial behavior, we cannot make conclusions about other subtypes (e.g., helping, comforting). In addition, separating the emotion recognition task from the prosocial behavior task limits and our cross-sectional design our ability to make conclusions about directionality of these mechanism. Potential improvements to our protocol are discussed further below.

Implications and Future Directions

This study was the first that we know of to examine the empathic behavioral outcomes of eye gaze, CU traits, and emotion recognition. While our results were similar to prior studies

(e.g., Williams & Moore, 2016), relationships between CU traits and prosocial behavior were minimal, at best, in contrast to the view that CU traits reflect a lack of prosocial emotions (DSM-

5; APA, 2013). This discrepancy suggests that this prosocial behavior task may be different from prosocial behaviors observed in children’s everyday social and school settings (e.g., Belacchi, &

Farina, 2012) and that perception of equality may play a role (Shaw et al., 2016). It is possible that naturalistic settings include additional social cues that would alter prosocial responding, such as context or vocal and postural communication, which were absent in the paradigm we selected.

34

Future studies should strive to include such social information in their paradigms as well as creating a paradigm that could test eye gaze, emotion recognition, and prosocial responding with the same stimulus. This would allow for testing of mediation. Further, longitudinal studies can test the role of CU traits in emotion recognition and prosocial behavior throughout development, focusing on identifying if these are indeed causal relationships. In addition, future studies might examine distinct facets of CU traits, as well as isolating the impact of empathy.

As accuracy for fear recognition was significantly worse for those who initially fixated on the mouth, results suggest that emotion recognition training programs that call attention to the eye regions might be helpful for improving recognition of high valence fear expressions

(Schönenberg et al., 2014), yet they may be less effective for subtle expressions. There is some preliminary evidence that implicit emotion recognition training for low intensity expressions increases overall bias for detecting the trained emotions, however it does not increase recognition accuracy (Griffiths, Jarrold, Penton-Voak & Munafò, 2015). On the other hand, an explicit emotion recognition training task using subtle emotions did show significant increase in emotion recognition and decreases in severity of subsequent crimes (Hubble, Bowen, Moore & Van

Goozen, 2015). Clearly there are promising new interventions on the horizon and investigation of the differences in design is necessary in order to improve the likelihood of success.

Conclusions

This study demonstrated that for girls, CU traits predict less initial fixation on the eyes which corresponds to poorer emotion recognition of low intensity happy expressions, which in turn corresponds to less prosocial sharing behavior. While findings are preliminary in terms of application of a new emotion recognition task in combination with a prosocial sharing task in young female and male children, this study provides insight that eye gaze patterns and emotion

35

recognition for positive emotions may mediate the negative relationship between CU traits and prosocial behavior in girls in this age range. This preliminary investigation demonstrated the feasibility of implementing a new emotion recognition paradigm with more standardized stimuli on a younger child sample, and suggests potential improvements for further testing these relationships and their temporal relationships, which ultimately could inform future interventions.

36

References

Achenbach, T. M., & Rescorla, L. (2001). ASEBA school-age forms & profiles. Burlington, VT:

Aseba.

Adolphs, R., Gosselin, F., Buchanan, T. W., Tranel, D., Schyns, P., & Damasio, A. R. (2005). A

mechanism for impaired fear recognition after amygdala damage. Nature, 433(7021), 68-

72.

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders

(5th ed.). Arlington, VA: American Psychiatric Publishing.

Amso, D., Haas, S., & Markant, J. (2014). An eye tracking investigation of developmental

change in bottom-up attention orienting to faces in cluttered natural scenes. PloS

one, 9(1), e85701.

Aspan, N., Vida, P., Gadoros, J., & Halasz, J. (2013). Conduct symptoms and emotion

recognition in adolescent boys with externalization problems. The Scientific World

Journal, 199(3), 1-6.

Barry, C. T., Frick, P. J., DeShazo, T. M., McCoy, M. G., Ellis, M., & Loney, B. R. (2000). The

importance of callous–unemotional traits for extending the concept of psychopathy to

children. Journal of Abnormal Psychology, 109, 335–340.

Barth, J. M., & Bastiani, A. (1997). A longitudinal study of emotion recognition and preschool

children's social behavior. Merrill-Palmer Quarterly 43(1), 107-128.

Bartz, J. A., Zaki, J., Bolger, N., & Ochsner, K. N. (2011). Social effects of oxytocin in humans:

context and person matter. Trends in cognitive sciences, 15(7), 301-309.

Belacchi, C., & Farina, E. (2012). Feeling and thinking of others: Affective and cognitive

empathy and emotion comprehension in prosocial/hostile preschoolers. Aggressive

behavior, 38(2), 150-165.

Benesch, C., Görtz-Dorten, A., Breuer, D., & Döpfner, M. (2014). Assessment of Callous-

Unemotional Traits in 6 to 12 Year-Old Children with Oppositional Defiant

Disorder/Conduct Disorder by Parent Ratings. Journal of Psychopathology and

Behavioral Assessment, 36(4), 519-529.

Best, C. A., Yim, H., & Sloutsky, V. M. (2013). The cost of selective attention in category

learning: Developmental differences between adults and infants. Journal of experimental

child psychology, 116(2), 105-119.

Blair, R. J. R. (1995). A cognitive developmental approach to morality: Investigating the

psychopath. Cognition, 57(1), 1-29.

Blair, R. J. R. (2003). Facial expressions, their communicatory functions and neuro–cognitive

substrates. Philosophical Transactions of the Royal Society B: Biological Sciences,

358(1431), 561-572.

Blair, R. J. R. (2006). The emergence of psychopathy: Implications for the neuropsychological

approach to developmental disorders. Cognition, 101, 414–442.

Blair, R. J. R., & Coles, M. (2000). Expression recognition and behavioural problems in early

adolescence. Cognitive development, 15(4), 421-434.

Blair, R. J., Colledge, E., Murray, L., & Mitchell, D. G. (2001). A selective impairment in the

processing of sad and fearful expressions in children with psychopathic tendencies.

Journal of Abnormal Child Psychology, 29(6), 491–8.

38

Blair, R. J. R., Mitchell, D. G. V., Peschardt, K. S., Colledge, E., Leonard, R. A., Shine, J. H., ...

& Perrett, D. I. (2004). Reduced sensitivity to others’ fearful expressions in psychopathic

individuals. Personality and Individual Differences, 37(6), 1111-1122.

Bowen, K. L., Morgan, J. E., Moore, S. C., & van Goozen, S. H. (2014). Young offenders’

emotion recognition dysfunction across emotion intensities: explaining variation using

psychopathic traits, conduct disorder and offense severity. Journal of psychopathology

and behavioral assessment, 36(1), 60-73.

Byrd, A. L., Kahn, R. E., & Pardini, D. A. (2013). A validation of the inventory of callous-

unemotional traits in a community sample of young adult males. Journal of

psychopathology and behavioral assessment, 35(1), 20-34.

Calder, A., Young, A., Perrett, D., Etcoff, N., & Rowland, D. (1996). Categorical perception of

morphed facial expressions. Visual Cognition, 3(2), 81-117.

http://doi.org/10.1080/713756735

Campbell, A. (2010). Oxytocin and human social behavior. Personality and Social Psychology

Review, 14(3) 281–295.

Cleckley, H. (1976). The mask of sanity 5th edn. St Louis: Mosby.

Colder, C. R., & O'Connor, R. M. (2004). Gray's reinforcement sensitivity model and child

psychopathology: Laboratory and questionnaire assessment of the BAS and BIS. Journal

of Abnormal Child Psychology,32(4), 435-451.

Colder, C. R., Trucco, E. M., Lopez, H. I., Hawk, L. W., Read, J. P., Lengua, L. J., ... & Eiden,

R. D. (2011). Revised reinforcement sensitivity theory and laboratory assessment of BIS

and BAS in children. Journal of Research in Personality, 45(2), 198-207.

39

Cooke, D. J., & Michie, C. (2001). Refining the construct of psychopathy: towards a hierarchical

model. Psychological assessment, 13(2), 171-188.

Dadds, M. R., Allen, J. L., Oliver, B. R., Faulkner, N., Legge, K., Moul, C., ... & Scott, S.

(2012). Love, eye contact and the developmental origins of empathy v. psychopathy. The

British Journal of Psychiatry, 200(3), 191-196.

Dadds, M. R., Hawes, D. & Merz, S. (2004) The UNSW Facial Emotion Task. Sydney:

University of New South Wales.

Dadds, M. R., Jambrak, J., Pasalich, D., Hawes, D. J., & Brennan, J. (2011). Impaired attention

to the eyes of attachment figures and the developmental origins of psychopathy. Journal

of Child Psychology and Psychiatry, 52(3), 238-245.

Dadds, M. R., El Masry, Y., Wimalaweera, S., & Guastella, A. J. (2008). Reduced eye gaze

explains “fear blindness” in childhood psychopathic traits. Journal of the American

Academy of Child & Adolescent Psychiatry, 47(4), 455-463.

Dadds, M. R., Perry, Y., Hawes, D. J., Merz, S., Riddell, A. C., Haines, D. J., …

Abeygunawardane, A. I. (2006). Attention to the eyes and fear-recognition deficits in

child psychopathy. The British Journal of Psychiatry : The Journal of Mental Science,

189, 280–281.

Dawel, A., O’Kearney, R., McKone, E., & Palermo, R. (2012). Not just fear and sadness: meta-

analytic evidence of pervasive emotion recognition deficits for facial and vocal

expressions in psychopathy. Neuroscience and Biobehavioral Reviews, 36(10), 2288–

2304.

Dawel, A., McKone, E., O’Kearney, R., Sellbom, M., Irons, J., & Palermo, R. (2015). Elevated

levels of callous unemotional traits are associated with reduced attentional cueing, with

40

no specificity for fear or eyes. Personality Disorders: Theory, Research, and

Treatment, 6(3), 216-228.

Dunfield, K. A. (2014). A construct divided: prosocial behavior as helping, sharing, and

comforting subtypes. Frontiers in Psychology, 5 (958), 111-123.

Edens, J. F., Marcus, D. K., Lilienfeld, S. O., & Poythress Jr, N. G. (2006). Psychopathic, not

psychopath: Taxometric evidence for the dimensional structure of psychopathy. Journal

of Abnormal Psychology, 115, 131-144.

Eisenbarth, H., Demetriou, C. A., Kyranides, M. N., & Fanti, K. A. (2016). Stability Subtypes of

Callous–Unemotional Traits and Conduct Disorder Symptoms and Their

Correlates. Journal of youth and adolescence, 45(9), 1889-1901.

Eisenberg, N., & Eggum, N. D. (2009). Empathic responding: Sympathy and personal

distress. In J. Decety & W. Ickes, The social neuroscience of empathy, (pp.71-83). MIT

Press.

Eisenberg, N., Eggum, N. D., & Di Giunta, L. (2010). Empathy related responding: Associations

with prosocial behavior, aggression, and intergroup relations. Social Issues and Policy

Review, 4, 143–180.

Eisenberg, N., Fabes, R. A., & Spinrad, T. L. (2006). Prosocial development. In N. Eisenberg &

W. Damon (Eds.), Handbook of child psychology. 6th edition (646-718). New York:

Wiley.

Eisenberg, N., Huerta, S., & Edwards, A. (2012). Relation of empathy-related responding to

children’s and adolescents’ social competence. Empathy: From bench to bedside, 148-

163.

41

Eisenberg, N., Spinrad, T. L., & Sadovsky, A. (2006). Empathy-related responding in

children. Handbook of moral development, 517-549.

Ezpeleta, L., Osa, N. D. L., Granero, R., Penelo, E., & Domènech, J. M. (2013). Inventory of

callous-unemotional traits in a community sample of preschoolers. Journal of Clinical

Child & Adolescent Psychology, 42(1), 91-105.

Fanti, K. A., & Centifanti, L. C. M. (2014). Childhood callous-unemotional traits moderate the

relation between parenting distress and conduct problems over time. Child Psychiatry &

Human Development, 45(2), 173-184.

Fanti, K. A., Panayiotou, G., Lazarou, C., Michael, R., & Georgiou, G. (2015). The better of two

evils? Evidence that children exhibiting continuous conduct problems high or low on

callous–unemotional traits score on opposite directions on physiological and behavioral

measures of fear. Development and psychopathology, 28, 185-198.

Fehr, E., Bernhard, H., & Rockenbach, B. (2008). Egalitarianism in young children. Nature, 454,

1079–1084.

Fehr, E., & Schmidt, K. (1999). A theory of fairness, competition, and cooperation. Quarterly

Journal of Economics, 114, 817–868.

Feshbach, N.D. (1978). Studies of empathic behaviour in children. Progress in experimental

personality research, 8, 1-47.

Feshbach, N. D., & Roe, K. (1968). Empathy in six-and seven-year-olds. Child development,

133-145.

Fontaine, N. M., McCrory, E. J., Boivin, M., Moffitt, T. E., & Viding, E. (2011). Predictors and

outcomes of joint trajectories of callous–unemotional traits and conduct problems in

childhood. Journal of abnormal psychology, 120(3), 730.

42

Fowles, D. C., & Kochanska, G. (2000). Temperament as a moderator of pathways to conscience

in children: The contribution of electrodermal activity. Psychophysiology, 37, 788-795.

Frank, M. G., Ekman, P., & Friesen, W. V. (1993). Behavioral markers and recognizability of the

smile of enjoyment. Journal of personality and social psychology, 64(1), 83-93.

Frick, P. J. (2004). The Inventory of Callous-Unemotional Traits. Unpublished rating scale.

Frick, P. J. (2009). Extending the construct of psychopathy to youth: Implications for

understanding, diagnosing, and treating antisocial children and adolescents. Canadian

Journal of Psychiatry/Revue canadienne de psychiatrie, 31(12), 803.

Frick, P. J. (2012). Developmental pathways to conduct disorder: Implications for future

directions in research, assessment, and treatment. Journal of Clinical Child & Adolescent

Psychology, 41(3), 378-389.

Frick, P. J., Blair, R. J., & Castellanos, F. X. (2013). Callous-unemotional traits and

developmental pathways to the disruptive behavior disorders. In Disruptive behavior

disorders (69-102). New York: Springer.

Frick, P. J., & Ellis, M. (1999). Callous–unemotional traits and subtypes of conduct disorder.

Clinical Child and Family Psychology Review, 2, 149–168.

Frick, P., & Hare, R. (2002). The Antisocial Process Screening Device. Toronto: Multi-Health

Systems.

Frick, P. J., Kimonis, E. R., Dandreaux, D. M., & Farell, J. M. (2003). The 4 year stability of

psychopathic traits in non-referred youth. Behavioral Science and the Law, 21, 713–736.

Frick, P. J., & Ray, J. V. (2014). Evaluating Callous‐Unemotional Traits as a Personality

Construct. Journal of personality. DOI: 10.1111/jopy.12114

43

Frick, P. J., Ray, J. V., Thornton, L. C., & Kahn, R. E. (2014). Annual Research Review: A

developmental psychopathology approach to understanding callous‐unemotional traits in

children and adolescents with serious conduct problems. Journal of child Psychology and

Psychiatry, 55(6), 532-548.

Fritz, M. S., & MacKinnon, D. P. (2007). Required sample size to detect the mediated

effect. Psychological science, 18(3), 233-239.

Gao, X., Maurer, D., & Nishimura, M. (2010). Similarities and differences in the perceptual

structure of facial expressions of children and adults. Journal of Experimental Child

Psychology, 105(1-2), 98–115. http://doi.org/10.1016/j.jecp.2009.09.001

Gillespie, S. M., Rotshtein, P., Satherley, R.-M., Beech, A. R., & Mitchell, I. J. (2015).

Emotional expression recognition and attribution bias among sexual and violent

offenders: a signal detection analysis. Frontiers in Psychology, 6(May), 1–12.

http://doi.org/10.3389/fpsyg.2015.00595

Griffiths, S., Jarrold, C., Penton-Voak, I. S., & Munafò, M. R. (2015). Feedback training induces

a bias for detecting happiness or fear in facial expressions that generalises to a novel task.

Psychiatry Research, 230(3), 951-957.

Guyer, A. E., Monk, C. S., McClure-Tone, E. B., Nelson, E. E., Roberson-Nay, R.,... & Ernst, M.

(2008). A developmental examination of amygdala response to facial

expressions. Journal of Cognitive Neuroscience, 20(9), 1565-1582.

Han, T., Alders, G. L., Greening, S. G., Neufeld, R. W. J., & Mitchell, D. G. V. (2012). Do

fearful eyes activate empathy-related brain regions in individuals with callous traits?

Social Cognitive and Affective Neuroscience, 7(8), 958–68.

44

Hare, R. D., Harpur, T. J., Hakstian, A. R., Forth, A. E., Hart, S. D., & Newman, J. P. (1990).

The revised Psychopathy Checklist: reliability and factor structure. Psychological

Assessment: A Journal of Consulting and Clinical Psychology, 2(3), 338.

Hare, R. D., Williamson, S. E., & Harpur, T. J. (1988). Psychopathy and language. In T. E.

Moffitt, & S. A. Mednick, Biological Contributions to Crime Causation (68-92).

Dordrecht, Netherlands: Martinus Nijhoff.

Harter, S. (1983). Children's understanding of multiple emotions: A cognitive-developmental

approach. In Overton, W. S. (Ed.), The relationship between social and cognitive

development, 147-194. Psychology Press.

Hawes, S. W., Byrd, A. L., Henderson, C. E., Gazda, R. L., Burke, J. D., Loeber, R., & Pardini,

D. A. (2014). Refining the parent-reported Inventory of Callous–Unemotional Traits in

boys with conduct problems. Psychological assessment, 26(1), 256.

Hawes, D. J., & Dadds, M. R. (2007). Stability and malleability of callous-unemotional traits

during treatment for childhood conduct problems. Journal of Clinical Child and

Adolescent Psychology, 36(3), 347-355.

Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis: A

regression-based approach. Guilford Press.

Herpers, P. C. M., Scheepers, F. E., Bons, D. M. a, Buitelaar, J. K., & Rommelse, N. N. J.

(2014). The cognitive and neural correlates of psychopathy and especially callous-

unemotional traits in youths: a systematic review of the evidence. Development and

Psychopathology, 26(1), 245–73.

Heyman, G., Barner, D., Heumann, J., & Schenck, L. (2014). Children's sensitivity to ulterior

motives when evaluating prosocial behavior. Cognitive science, 38(4), 683-700.

45

Hicks, B. M., Carlson, M. D., Blonigen, D. M., Patrick, C. J., Iacono, W. G., & Mgue, M.

(2012). Psychopathic personality traits and environmental contexts: Differential

correlates, gender differences, and genetic mediation. Personality Disorders, 3(3), 209–

27.

Houghton, S., Hunter, S. C., & Crow, J. (2013). Assessing callous unemotional traits in children

aged 7-to 12-years: A confirmatory factor analysis of the inventory of callous

unemotional traits. Journal of Psychopathology and Behavioral Assessment, 35(2), 215-

222.

Hubble, K., Bowen, K. L., Moore, S. C., & Van Goozen, S. H. (2015). Improving negative

emotion recognition in young offenders reduces subsequent crime. PloS one, 10(6),

e0132035.

Hunnius, S., de Wit, T. C., Vrins, S., & von Hofsten, C. (2011). Facing threat: Infants' and adults'

visual scanning of faces with neutral, happy, sad, angry, and fearful emotional

expressions. Cognition and emotion, 25(2), 193-205.

Kahn, R. E., Frick, P. J., Youngstrom, E. A, Youngstrom, J. K., Feeny, N. C., & Findling, R. L.

(2013). Distinguishing primary and secondary variants of callous-unemotional traits among

adolescents in a clinic-referred sample. Psychological Assessment, 25(3), 966–78.

http://doi.org/10.1037/a0032880

Kimonis, E. R., Fanti, K. A., & Singh, J. P. (2014). Establishing Cut-off Scores for the Parent-

Reported Inventory of Callous-Unemotional Traits. Archives of Forensic

Psychology, 1(1), 27-48.

Kimonis, E. R., Fanti, K. A., Frick, P. J., Moffitt, T. E., Essau, C., Bijttebier, P., & Marsee, M.

A. (2015). Using self‐reported callous‐unemotional traits to cross‐nationally assess the

46

DSM‐5 ‘With Limited Prosocial Emotions’ specifier. Journal of Child Psychology and

Psychiatry, 56(11), 1249-1261.

Kochanska, G. (1993). Toward a synthesis of parental socialization and child temperament in

early development of conscience. Child development, 64(2), 325-347.

Kochanska, G., & Aksan, N. (2006). Children's conscience and self‐regulation. Journal of

Personality, 74(6), 1587-1618.

Lawrence, K., Campbell, R., & Skuse, D. (2015). Age, gender, and puberty influence the

development of facial emotion recognition. Frontiers in psychology, 6(761), 1-14.

Leist, T., & Dadds, M. R. (2009). Adolescents' ability to read different emotional faces relates to

their history of maltreatment and type of psychopathology. Clinical Child Psychology

and Psychiatry, 14(2), 237-250.

Leitzke, B. T., & Pollak, S. D. (2016). Developmental changes in the primacy of facial cues for

emotion recognition. Developmental psychology, 52(4), 572.

Leppänen, J. M., & Hietanen, J. K. (2001). Emotion recognition and social adjustment in school–

aged girls and boys. Scandinavian journal of psychology,42(5), 429-435.

LoBue, V., Nishida, T., Chiong, C., DeLoache, J. S., & Haidt, J. (2011). When getting something

good is bad: Even three‐year‐olds react to inequality. Social Development, 20(1), 154-

170.

López-Romero, L., Romero, E., & Villar, P. (2014). Assessing the stability of psychopathic

traits: adolescent outcomes in a six-year follow-up. The Spanish journal of

psychology, 17, E97.

Lozier, L. M., Cardinale, E. M., VanMeter, J. W., & Marsh, A. a. (2014). Mediation of the

relationship between callous-unemotional traits and proactive aggression by amygdala

47

response to fear among children with conduct problems. JAMA Psychiatry, 71(6), 627–

36.

Luman, M., van Meel, C. S., Oosterlaan, J., & Geurts, H. M. (2012). Reward and punishment

sensitivity in children with ADHD: validating the sensitivity to punishment and

sensitivity to reward questionnaire for children (SPSRQ-C).Journal of abnormal child

psychology, 40(1), 145-157.

Lykken, D. T. (1995). The antisocial personalities. Psychology Press.

Malti, T., Gummerum, M., Keller, M., & Buchmann, M. (2009). Children’s moral motivation,

sympathy, and prosocial behavior. Child development, 80(2), 442-460.

Malti, T., & Krettenauer, T. (2013). The Relation of Moral Emotion Attributions to Prosocial and

Antisocial Behavior: A Meta‐Analysis. Child development, 84(2), 397-412.

Marsh, A. A., Finger, E. C., Mitchell, D. G. V., Reid, M., Sims, C., Kosson, D. S., et al. (2008).

Reduced amygdala response to fearful expressions in children and adolescents with

callous–unemotional traits and disruptive behaviour disorders. American Journal of

Psychiatry, 165, 712–720.

Marsh, A. a., Finger, E. C., Schechter, J. C., Jurkowitz, I. T. N., Reid, M. E., & Blair, R. J. R.

(2011). Adolescents with psychopathic traits report reductions in physiological responses

to fear. Journal of Child Psychology and Psychiatry and Allied Disciplines, 52, 834–841.

http://doi.org/10.1111/j.1469-7610.2010.02353.x

Marsh, A. A., Henry, H. Y., Pine, D. S., & Blair, R. J. R. (2010). Oxytocin improves specific

recognition of positive facial expressions. Psychopharmacology, 209(3), 225-232.

Matthews, G. & Wells, A. 1999 The cognitive science of attention and emotion. In Handbook of

cognition and emotion (ed. T. Dalgleish & M. J. Power), pp. 171–192. New York: Wiley.

48

Michalska, K. J., Kinzler, K. D., & Decety, J. (2013). Age-related sex differences in explicit

measures of empathy do not predict brain responses across childhood and

adolescence. Developmental cognitive neuroscience, 3, 22-32.

Miller, A. L., Gouley, K. K., Seifer, R., Zakriski, A., Eguia, M., & Vergnani, M. (2005).

Emotion Knowledge Skills in Low‐income Elementary School Children: Associations

with Social Status and Peer Experiences. Social Development,14(4), 637-651.

Mizokawa, A., Minemoto, K., Komiya, A., & Noguchi, M. (2013). The Effects of Infants’ and

Adults’ Facial Expressions on Approach-Avoidance Behavior. Psychologia, 56(1), 33-44.

Moffitt, T. E. (1993). Adolescence-limited and life-course-persistent antisocial behavior: a

developmental taxonomy. Psychological review, 100(4), 674.

Muñoz, L. C. (2009). Callous-unemotional traits are related to combined deficits in recognizing

afraid faces and body poses. Journal of the American Academy of Child & Adolescent

Psychiatry, 48(5), 554-562.

Muratori, P., Lochman, J. E., Manfredi, A., Milone, A., Nocentini, A., Pisano, S., & Masi, G.

(2016). Callous unemotional traits in children with disruptive behavior disorder:

predictors of developmental trajectories and adolescent outcomes. Psychiatry

research, 236, 35-41.

Murray-Close, D., & Crick, N. R. (2007). Gender differences in the association between

cardiovascular reactivity and aggressive conduct. International Journal of

Psychophysiology : Official Journal of the International Organization of

Psychophysiology, 65(2), 103–13.

49

Paiva-Salisbury, M. L., Gill, A. D., & Stickle, T. R. (2016). Isolating Trait and Method Variance

in the Measurement of Callous and Unemotional Traits. Assessment,

doi:10.1177/1073191115624546.

Paulus, M., Kühn-Popp, N., Licata, M., Sodian, B., & Meinhardt, J. (2013). Neural correlates of

prosocial behavior in infancy: different neurophysiological mechanisms support the

emergence of helping and comforting. Neuroimage, 66, 522-530.

Pardini, D., & Frick, P. J. (2013). Multiple developmental pathways to conduct disorder: Current

conceptualizations and clinical implications. Journal of the Canadian Academy of Child

and Adolescent Psychiatry, 22(1), 20.

Pardini, D. A., Lochman, J. E., & Frick, P. J. (2003). Callous/unemotional traits and social-

cognitive processes in adjudicated youths. Journal of the American Academy of Child &

Adolescent Psychiatry, 42(3), 364-371.

Patrick, C. J., Bradley, M. M., & Lang, P. J. (1993). Emotion in the Criminal Psychopath: Startle

Reflex Modulation. Journal of Abnormal Psychology, 82-92.

Ray, J. V., Frick, P. J., Thornton, L. C., Steinberg, L., & Cauffman, E. (2016). Positive and

negative item wording and its influence on the assessment of callous-unemotional traits.

Psychological Assessment, 28(4), 394-404.

Rosenqvist, J., Lahti-Nuuttila, P., Laasonen, M., & Korkman, M. (2014). Preschoolers’

recognition of emotional expressions: Relationships with other neurocognitive

capacities. Child Neuropsychology, 20(3), 281-302.

Salihovic, S., Kerr, M., & Stattin, H. (2014). Under the surface of adolescent psychopathic traits:

high-anxious and low-anxious subgroups in a community sample of youths. Journal of

Adolescence, 37(5), 681–9. http://doi.org/10.1016/j.adolescence.2014.03.002

50

Schönenberg, M., Christian, S., Gaußer, a-K., Mayer, S. V, Hautzinger, M., & Jusyte, a. (2014).

Addressing perceptual insensitivity to facial affect in violent offenders: first evidence for

the efficacy of a novel implicit training approach. Psychological Medicine, 44(5), 1043–

52.

Schönenberg, M., Mayer, S. V., Christian, S., Louis, K., & Jusyte, A. (2015). Facial affect

recognition in violent and nonviolent antisocial behavior subtypes. Journal of Personality

Disorders, 29, 1-12.

Seidela, E. M., Habela, U., Kirschnerc, M., Gurd, R. C., & Derntl, B. (2010). The impact of

facial emotional expressions on behavioral tendencies in females and males. J Exp

Psychol Hum Percept Perform, 36(2), 500-507.

Sharp, C., van Goozen, S., & Goodyer, I. (2006). Children’s subjective emotional reactivity to

affective pictures: Gender differences and their antisocial correlates in an unselected

sample of 7-11-year-olds. Journal of Child Psychology and Psychiatry and Allied

Disciplines, 47, 143–150.

Sharp, C. (2008). Theory of Mind and conduct problems in children: Deficits in reading the

“emotions of the eyes”. Cognition and Emotion, 22(6), 1149-1158.

Shaw, A., Choshen-Hillel, S., & Caruso, E. M. (2016). The development of inequity aversion

understanding when (and why) people give others the bigger piece of the

pie. Psychological Science, 0956797616660548.

Sheskin, M., Bloom, P., & Wynn, K. (2014). Anti-equality: Social comparison in young

children. Cognition, 130(2), 152-156.

51

Spezio, M. L., Huang, P.-Y. S., Castelli, F., & Adolphs, R. (2007). Amygdala damage impairs

eye contact during conversations with real people. The Journal of Neuroscience : The

Official Journal of the Society for Neuroscience, 27(15), 3994–3997.

Steiger, J. H. (1980). Tests for comparing elements of a correlation matrix. Psychological

Bulletin, 87, 245-251.

Stevens, D., Charman, T., & Blair, R. J. R. (2001). Recognition of emotion in facial expressions

and vocal tones in children with psychopathic tendencies. The Journal of genetic

psychology, 162(2), 201-211.

Stuijfzand, S., De Wied, M., Kempes, M., Van de Graaff, J., Branje, S., & Meeus, W. (2016).

Gender Differences in Empathic Sadness towards Persons of the Same-versus Other-sex

during Adolescence. Sex Roles, 1-13.

Székely, E., Tiemeier, H., Jaddoe, V. W., Hofman, A., Verhulst, F. C., & Herba, C. M. (2014).

Associations of Internalizing and Externalizing Problems with Facial Expression

Recognition in Preschoolers: The Generation R Study. Social Development, 23(3), 611-

630.

Tell, D., Davidson, D., & Camras, L. a. (2014). Recognition of emotion from facial expressions

with direct or averted eye gaze and varying expression intensities in children with autism

disorder and typically developing children. Autism Research and Treatment, 2014, 1-11.

http://doi.org/10.1155/2014/816137

Thoemmes, F., MacKinnon, D. P., & Reiser, M. R. (2010). Power analysis for complex

mediational designs using Monte Carlo methods. Structural Equation Modeling, 17(3),

510-534.

52

Tottenham, N., Tanaka, J. W., Leon, A. C., McCarry, T., Nurse, M., Hare, T. A., ... & Nelson, C.

(2009). The NimStim set of facial expressions: judgments from untrained research

participants. Psychiatry research, 168(3), 242-249.

Viding, E., Frick, P. J., & Plomin, R. (2007). Aetiology of the relationship between callous–

unemotional traits and conduct problems in childhood. The British Journal of

Psychiatry, 190(49), 33-38.

Viding, E., Sebastian, C. L., Dadds, M. R., Lockwood, P. L., Cecil, C. a M., De Brito, S. a, &

McCrory, E. J. (2012). Amygdala response to preattentive masked fear in children with

conduct problems: the role of callous-unemotional traits. The American Journal of

Psychiatry, 169(10), 1109–16.

Waller, R., Gardner, F., & Hyde, L. W. (2013). What are the associations between parenting,

callous–unemotional traits, and antisocial behavior in youth? A systematic review of

evidence. Clinical Psychology Review, 33(4), 593-608.

Waller, R., Wright, A. G., Shaw, D. S., Gardner, F., Dishion, T. J., Wilson, M. N., & Hyde, L.

W. (2015). Factor structure and construct validity of the parent-reported inventory of

callous-unemotional traits among high-risk 9-year-olds. Assessment, 22(5), 561-580.

White, B. A. (2014). Who cares when nobody is watching? Psychopathic traits and empathy in

prosocial behaviors. Personality and Individual Differences, 56, 116-121.

Williams, A., O’Driscoll, K., & Moore, C. (2014). The influence of empathic concern on

prosocial behavior in children. Frontiers in Psychology, 5, 1–8.

Williams, A., & Moore, C. (2016). A longitudinal exploration of advantageous and

disadvantageous inequality aversion in children. Journal of Experimental Child

Psychology, 152, 294-306.

53

Willoughby, M. T., Mills-Koonce, W. R., Waschbusch, D. A., & Gottfredson, N. C. (2015). An

examination of the parent report version of the inventory of callous-unemotional traits in

a community sample of first-grade children. Assessment, 22(1), 76-85.

Zajdel, R. T., Bloom, J. M., Fireman, G., & Larsen, J. T. (2013). Children's understanding and

experience of mixed emotions: The roles of age, gender, and empathy. The Journal of

genetic psychology, 174(5), 582-603.

54

Table 1 Trial Types for Prosocial Behavior Task Child, Jenny vs. Child, Jenny AI No Cost- 4x 1, 0 1, 1 AI Cost- 4x 2, 0 1, 1 DI No Cost- 4x 1, 2 1, 1 DI Cost- 4x 2 ,3 1, 1 Note: AI = Advantageous Inequality, DI = Disadvantageous Inequality

Table 2 Descriptive Statistics for Primary Variables and Accuracy Std S.E. S.E. Variable Mean Dev Kurtosis Kurt Skew Skew Min Max Sex .58 .50 -1.94 .53 -.33 .27 .00 1.0 Age (months) 93.84 13.83 -1.28 .53 -.04 .27 72.00 119.0 AI 7.11 1.36 1.03 .53 -1.49 .27 3.04 8.0 DI 3.73 2.80 -1.31 .53 .02 .27 .00 8.0 AI No Cost 3.72 .58 6.28 .53 -2.33 .27 1.00 4.0 AI Cost 3.40 1.00 1.53 .53 -1.57 .27 .00 4.0 DI No Cost 1.54 1.51 -1.29 .53 .42 .27 .00 4.0 DI Cost 2.19 1.64 -1.58 .53 -.27 .27 .00 4.0 ICU Total 16.88 8.06 1.32 .53 1.03 .27 3.00 41.0 Fearfulness 19.25 5.66 -.13 .53 .22 .27 8.00 36.0 Anxiety Problems 54.61 7.30 1.35 .53 1.58 .27 50.00 76.8 Accuracy at 50% Fear .62 .27 -.31 .53 -.60 .27 .00 1.0 Happy .74 .23 .01 .53 -.79 .27 .04 1.0 Sad .51 .24 -.74 .53 .18 .27 .00 1.0 Angry .61 .18 .59 .53 -.73 .27 .05 1.0 Accuracy at 100% Fear .77 .25 1.06 .53 -1.22 .27 .01 1.0 Happy .93 .13 2.56 .53 -1.82 .27 .50 1.0 Sad .77 .24 .34 .53 -1.04 .27 .04 1.0 Angry .89 .15 1.22 .53 -1.37 .27 .41 1.0 Note. Sex: 0 = male, 1 = female, Race: 0 = White, 1 = other, ICU total = Inventory of Callous-Unemotional Traits Total Score, Fear/Shyness = Fear/Shyness scale of the SPSRQ-C, Anxiety Problems = Anxiety Problems Subscale of the CBCL, AI = Advantageous Inequality, DI = Disadvantageous Inequality

55

Table 3. Descriptive Statistics of Initial Fixation Variable Mean Std Dev Kurtosis S.E. Kurt Skew S.E. Skew Min Max Proportion of Initial Fixation on the Eyes vs. Elsewhere on Face at 50% Stimuli Intensity Fear .63 .24 -.06 .53 -.49 .27 0 1.00 Happy .60 .23 -.49 .53 -.28 .27 0 1.00 Sad .63 .26 -.56 .53 -.37 .27 0 1.00 Angry .62 .24 -.26 .53 -.10 .27 0 1.00 Proportion of Initial Fixation on the Eyes vs. Elsewhere on Face at 100% Stimuli Intensity Fear .59 .27 -.93 .53 -.07 .27 0 1.00 Happy .61 .24 -.42 .53 -.28 .27 0 1.00 Sad .63 .24 .04 .53 -.47 .27 0 1.00 Angry .62 .25 -.29 .53 -.51 .27 0 1.00 Proportion of Initial Fixation on the Mouth vs. Elsewhere on Face at 50% Stimuli Intensity Fear .14 .18 .79 .53 1.29 .27 0 .67 Happy .16 .20 1.54 .53 1.37 .27 0 .80 Sad .15 .18 .52 .53 1.11 .27 0 .67 Angry .17 .20 2.00 .53 1.26 .27 0 1.00 Proportion of Initial Fixation on the Mouth vs. Elsewhere on Face at 100% Stimuli Intensity Fear .22 .23 .07 .53 .99 .27 0 .83 Happy .16 .20 1.41 .53 1.30 .27 0 .79 Sad .17 .22 .51 .53 1.18 .27 0 .83 Angry .16 .21 1.26 .53 1.42 .27 0 .78

56

Table 4. Descriptive Statistics of Gaze Duration Std S.E. S.E. Variable Mean Dev Kurtosis Kurt Skewness Skew Min Max Duration of Gaze on the Eyes at 50% Fear 858.61 250.41 .49 .53 .17 .27 94.80 1464.18 Happy 751.72 252.40 .43 .53 .25 .27 .00 1414.17 Sad 824.43 253.23 -.04 .53 .52 .27 280.61 1453.07 Angry 865.44 262.06 -.13 .53 .56 .27 366.74 1522.53 Duration of Gaze on the Eyes at 100% Fear 842.09 235.33 1.32 .53 .89 .27 316.73 1549.50 Happy 872.53 277.15 .47 .53 .43 .27 266.72 1715.40 Sad 862.20 279.18 .54 .53 -.06 .27 21.90 1520.30 Angry 882.88 247.13 -.01 .53 .58 .27 325.07 1525.31 Duration of Gaze on the Mouth at 50% Fear 544.97 206.92 .78 .53 .44 .27 .00 1186.30 Happy 640.63 238.93 .54 .53 -.02 .27 .00 1300.26 Sad 608.66 224.30 .88 .53 .97 .27 191.71 1303.20 Angry 567.09 220.92 -.18 .53 .37 .27 150.03 1186.35 Duration of Gaze on the Mouth at 100% Fear 620.66 217.08 .57 .53 .26 .27 180.04 1289.70 Happy 565.93 255.96 .37 .53 .59 .27 .00 1347.80 Sad 565.39 260.02 .85 .53 .69 .27 .00 1330.82 Angry 566.14 202.67 .58 .53 .72 .27 244.49 1180.30

57

Table 5. Paired Samples T tests for Emotion Recognition Paired Differences 95% Confidence Interval of the Difference Std. Std. Error Sig (2- Pairs Mean Dev. mean Lower Upper T df tailed) 50% Emotion Intensity Fear-Happy -.13 .30 .03 -.20 -.06 -3.83 80 <.001 Fear-Sad .11 .33 .04 .03 .18 2.93 80 .004 Fear-Angry .00 .26 .03 -.06 .06 0.04 80 .969 Happy-Sad .24 .31 .03 .17 .31 6.99 80 <.001 Happy-Angry .13 .24 .03 .08 .18 4.88 80 <.001 Sad-Angry -.11 .29 .03 -.17 -.04 -3.37 80 .001 100% Emotion Intensity Fear-Happy -.16 .27 .03 -.22 -.10 -5.31 80 <.001 Fear-Sad .00 .30 .03 -.07 .06 -0.14 80 .888 Fear-Angry -.12 .22 .02 -.17 -.07 -4.70 80 <.001 Happy-Sad .15 .25 .03 .10 .21 5.41 80 <.001 Happy-Angry .04 .20 .02 .00 .08 1.78 80 .080 Sad-Angry -.11 .29 .03 -.18 -.05 -3.47 80 .001

Table 6. Paired Samples T tests for Prosocial Behavior Task Paired Differences 95% Confidence Interval of the Difference Std. Std. Error Sig (2- Pairs Mean Dev. mean Lower Upper T df tailed) AI-DI 3.38 3.24 .36 2.67 4.10 9.39 80 <.001 AI Cost vs. .33 .86 .10 .14 .52 3.44 80 .001 No Cost DI Cost vs. -.64 1.44 .16 -.96 -.32 -4.00 80 <.001 No Cost Note. AI = Advantageous Inequality, DI = Disadvantageous Inequality

Table 7. Bivariate Correlations Age ICU Fear/ Anxiety AI No DI No Sex (months) Race total Shyness Problems Cost AI Cost Cost DI Cost Sex 1.00 .12 -.02 -.04 .19* .15 .19* .04 .09 .11 Age (months) .12 1.00 -.21* -.09 -.07 .11 -.10 .11 .36*** .32*** Race -.02 -.21* 1.00 .06 .13 .05 -.07 .00 .10 .11 ICU total -.04 -.09 .06 1.00 .03 .24** -.09 -.11 -.06 -.11 Fear/Shyness .19* -.07 .13 .03 1.00 .32*** .19* .12 -.15 .05 Anxiety Problems .15 .11 .05 .24** .32*** 1.00 .07 .21* .02 -.11 AI No Cost .19* -.10 -.07 -.09 .19* .07 1.00 .46*** -.14 .02 AI Cost .04 .11 .00 -.11 .12 .21* .46*** 1.00 -.04 -.16 DI No Cost .09 .36*** .10 -.06 -.15 .02 -.14 -.04 1.00 .58*** DI Cost .11 .32*** .11 -.11 .05 -.11 .02 -.16 .58*** 1.00 Note. Sex: 0 = male, 1 = female, Race: 0 = White, 1 = other, ICU total = Inventory of Callous-Unemotional Traits Total Score, Fear/Shyness = Fear/Shyness scale of the SPSRQ-C, Anxiety Problems = Anxiety Problems Subscale of the CBCL, AI = Advantageous Inequality, DI = Disadvantageous Inequality *p<.1, **p<.05, ***p<.01

Table 8. Correlations with Accuracy Age AI No DI No Sex (Months) Race Cost AI Cost Cost DI Cost ICU total Fear/shy Anx. Prob. Accuracy of Emotion Recognition at 50% Fear -.06 .20* .02 -.01 .00 .05 .00 -.17 -.15 -.15 Happy .06 .18 .07 -.12 .07 .21* .01 -.19* -.12 -.17 Sad .03 .19* .12 -.07 .18 .22* .15 -.11 .02 .01 Angry -.03 -.06 .02 -.06 -.12 .04 .04 -.09 -.27** -.14 Accuracy of Emotion Recognition at 100% Fear -.02 .11 -.07 .06 -.02 .05 .03 -.12 -.05 -.08 Happy .14 .20* .07 -.10 .03 .21* .17 -.28** .04 .01 Sad .13 .21* -.15 .03 .02 .10 .09 -.19* -.06 -.15 Angry -.11 .10 .04 -.03 -.13 .06 .13 .00 -.14 .03 Note. Sex: 0 = male, 1 = female, Race: 0 = White, 1 = other, ICU total = Inventory of Callous-Unemotional Traits Total Score, Fear/Shyness = Fear/Shyness scale of the SPSRQ-C, Anxiety Problems = Anxiety Problems Subscale of the CBCL, AI = Advantageous Inequality, DI = Disadvantageous Inequality *p<.1, **p<.05, ***p<.01

60

Table 9. Correlations with Initial Fixation Age AI No DI No Anxiety Emotion Sex (months) Race Cost AI Cost Cost DI Cost ICU total Fear/shy Problems Accuracy First Fixation on the Eyes at 50% Intensity Fear .07 .16 .05 -.03 .06 -.06 -.04 .02 -.12 -.17 .14 Happy .36*** .09 .05 .19* .11 .04 .13 -.23** .05 -.09 .27** Sad -.01 .32*** -.11 -.06 .13 .09 .02 -.25** -.12 .10 -.01 Angry .19* .31*** -.12 .10 .08 .25** .18 -.16 -.03 -.16 -.09 First Fixation on the Eyes at 100% Intensity Fear .06 .15 -.04 -.18 -.03 .15 .00 .00 .00 .01 .20* Happy .00 .03 .04 -.16 -.01 .06 -.07 .00 -.07 -.09 .15 Sad .06 .14 .05 -.05 .08 .07 .10 -.12 -.11 -.16 .00 Angry -.09 .15 .01 -.01 .06 .24** .12 -.10 -.10 -.16 .09 First Fixation on the Mouth at 50% Intensity Fear -.07 -.28** -.07 .02 -.08 -.16 -.10 -.01 .19* .13 -.20* Happy -.27** -.09 -.03 -.11 -.02 -.18 -.25** .13 -.03 .12 -.32*** Sad -.17 -.23** -.02 .01 -.16 -.14 -.12 .08 -.07 -.13 -.06 Angry -.19* -.28** .03 -.11 -.10 -.07 -.15 .19* .05 .12 -.04 First Fixation on the Mouth at 100% Intensity Fear -.06 -.12 -.01 .13 .10 -.13 -.10 -.07 .08 .12 -.38*** Happy -.12 -.05 -.11 .07 .01 -.14 -.01 .08 .01 .02 -.04 Sad -.03 -.17 .00 .05 -.05 -.08 -.07 .10 .09 .12 -.03 Angry -.07 -.29*** .04 .03 -.05 -.21* -.22** .15 .14 .18 -.19* Note. Sex: 0 = male, 1 = female, Race: 0 = White, 1 = other, ICU total = Inventory of Callous-Unemotional Traits Total Score, Fear/Shyness = Fear/Shyness scale of the SPSRQ-C, Anxiety Problems = Anxiety Problems Subscale of the CBCL, AI = Advantageous Inequality, DI = Disadvantageous Inequality *p<.1, **p<.05, ***p<.01

61

Table 10. Correlations with Duration Age AI No DI No Emotion Sex (months) Race Cost AI Cost Cost DI Cost ICU total Fear/shy Anx. Prob Accuracy Duration of Gaze on the Eyes at 50% Intensity Fear .19* .04 -.02 .05 .03 .12 .02 -.25** .08 -.11 .16 Happy .13 .02 .09 .18 .13 .19* .02 -.15 .05 -.13 .21* Sad .18 .16 -.08 .11 .13 .17 .01 -.25** -.06 -.16 .01 Angry .16 .04 .05 .08 .04 .11 .00 -.13 .05 -.10 .13 Duration of Gaze on the Eyes at 100% Intensity Fear .08 .10 -.01 -.05 .07 .13 -.14 -.12 -.05 -.08 .25** Happy .23** .18 -.01 -.04 .07 .14 .03 -.18 .06 -.21* .15 Sad .10 .01 -.12 .12 .12 .06 -.14 -.06 .16 -.07 -.02 Angry .08 .13 .08 .10 .06 .18 .06 -.15 .07 -.20* -.17 Duration of Gaze on the Mouth at 50% Intensity Fear .05 -.04 .03 .16 .12 -.04 -.04 .04 .17 .12 -.16 Happy .10 .02 -.02 .11 .14 .01 .05 -.01 .02 .31*** -.13 Sad .06 -.04 -.06 .13 .10 .00 -.04 .18 .02 .23** .05 Angry .13 .07 -.04 .05 .10 .06 .09 .06 -.01 .20* -.08 Duration of Gaze on the Mouth at 100% Intensity Fear .04 -.02 -.12 .10 .00 -.06 .06 -.05 .10 .22* -.27** Happy .00 .01 -.03 .00 -.01 -.07 -.01 -.08 .05 .18 .03 Sad .04 .04 -.10 .16 .05 -.03 .14 -.11 .05 .02 -.01 Angry .02 .07 -.18 -.12 .02 .03 .03 -.09 -.02 .25** -.08 Note. Sex: 0 = male, 1 = female, Race: 0 = White, 1 = other, ICU total = Inventory of Callous-Unemotional Traits Total Score, Fear/Shyness = Fear/Shyness scale of the SPSRQ-C, Anxiety Problems = Anxiety Problems Subscale of the CBCL, AI = Advantageous Inequality, DI = Disadvantageous Inequality *p<.1, **p<.05, ***p<.01

62

Table 11. Steiger Z tests for Significant ICU and Eye Gaze Correlations Correlation Correlation for Correlation for between 50% 50% intensity 100% intensity and 100% z-score 2-tail p ICU and Happy Initial Fixation -.23 .00 .19 -1.625 .104 ICU and Sad Initial Fixation -.25 -.12 .42 -1.094 .274 ICU and Fear Duration -.25 -.12 .64 -1.386 .166 ICU and Sad Duration -.25 -.06 .62 -1.967 .049 ICU and Fear Accuracy -.17 -.12 .52 -.457 .648 ICU and Sad Accuracy -.11 -.19 .29 .603 .547 ICU and happy Accuracy -.19 -.28 .15 .636 .525

63

Figure 1. Hypothesis 1

Figure 2. Hypothesis 2

Figure 3. Hypothesis 3

Figure 4. Hypothesis 4

Figure 5. Sample AOI Placement on NimStim image (Tottenham et al., 2009).

Figure 6. Moderated Mediation of CU traits on Recognition of Low Intensity Happy Expressions. Note: ap<.1, *p < .05

65

Figure 7. Interaction between Gender and Initial Fixation on the Eyes Predicting Accuracy of Recognition for Happy Faces at 50% intensity Males Females 0.9 0.8 b = .0032, p = .986 0.7 0.6 0.5 0.4 0.3 0.2 0.1 Faces at 50% at Intensity Faces 0.0 0.3709 0.6012 0.8316 Accuracy of Recognition Happy for Recognition of Accuracy Proportion First Fixating on the Eyes Compared to the Rest of the Face

Figure 8. Moderated Mediation of Eye Gaze on Prosocial Behavior. Note: ap<.1, *p < .05

66

Figure 9. Serial Mediation Model of CU Traits Predicting Prosocial Behavior Through Initial Fixation on the Eyes of Happy Faces at 50% Intensity. Note: ap<.1, *p < .05

67