Introduction

Prominent models of adult psychopathy define the disorder as the combination of antisocial behaviours (including criminal behaviours) and emotional coldness [28]. In children, the respective developmental antecedents of these factors are conduct problems (CP) and callous–unemotional traits (CU) [25]. CP samples are a heterogeneous group and the presence of high CU traits identifies a more homogenous subgroup (see [25]). High levels of CU traits in children are defined by a general lack of guilt, low concern for others, and limited emotional depth. The developmental course of these emotional impairments is not fully known, however, and there is some evidence that children with high CU traits may have the potential to be emotionally responsive under certain conditions. The current study examined responses to complex fear and attachment-related stimuli among children with high CU traits, in terms of their capacity to experience, and effectively regulate, genuine interpersonal emotions.

CU traits are evident and measurable by age four [12, 21], remain relatively stable through adolescence and into adulthood, and are associated with high bio-genetic influences [49]. Specifically, the impairments in emotional responsiveness and empathy are thought to have clear neurological correlates, and can be most clearly documented in laboratory settings whereby high CU children have impaired low responsiveness to, and inability to recognise, distress expressions, postures and auditory cues in others [2, 35]. We argue that there are four aspects of the putative emotion response/recognition impairment in high CU traits that are imprecisely defined and measured in the existing literature.

First, the majority of studies in this area have examined the ability to recognise distress from single cues, typically photographs [16], and have rarely used complex stimuli that involve more than still photos. Interestingly, in one of the few studies to use complex, emotional vignettes as their stimuli, it was found that high CU traits were not associated with poor recognition of emotions [51]. The current study builds on this by exposing children and their parent(s) to complex narratives of fear and attachment stimuli.

Second, stimuli are generally related to strangers rather than known people or caregivers, peers, and attachment figures. Emerging evidence indicates that parental warmth and involvement (relative to effective discipline and positive rewards) are especially important for protecting against conduct problems in children with high CU traits [32, 40]. This might indicate that children with high CU traits might be especially responsive to attachment figures early in life; however, relatively little is known about the influence of attachment relationships on emotion responsiveness and recognition in children with high CU traits. The few studies that have been conducted produce somewhat discrepant findings. Two studies have shown that children with CP and high CU traits are less affectionate and less attentive to the eye region of their caregivers during ‘love’ and free-interaction scenarios [9, 13]. Other evidence shows that CP children with high CU traits engage in relatively normal to high levels of emotional talk, including expression of self-referent fear language, with their caregivers [41, 42]. A major thrust of developmental research into CPs and psychopathy should be to track the developmental trajectories of cognitive and affective impairments in context, that is, in interaction with key attachment figures.

Third, the relationship of emotional responsiveness to attentional processes in high CU children is complex and interdependent [3, 17]. There is some evidence that adults with psychopathy and children with high CU traits can and do show emotional responsiveness and recognition when directed to pay attention to the emotion stimuli. A recent study found that adults with psychopathy show normal levels of fear recognition and startle responses when their attention is actively directed to the stimulus [37]. Similarly, directing the attention of high CU traits children to the eye region of emotion faces leads to better recognition of distress [11, 14]. It is likely that children with high CU traits have high thresholds for having their attention directed to emotional cues; that is, fewer emotional stimuli automatically capture attention. These thresholds may vary considerably according to the ‘attachment’ or other value of the context. Thus, it is possible that whereas emotional stimuli expressed by strangers (e.g., computerised photos) do not pass thresholds for attracting attention, those associated with attachment and other figures might, under certain conditions, have potential for eliciting emotional responses in high CU children.

One of the key facets of [9] seminal model of psychopathy is that these individuals are unable to form genuine emotional bonds with others. Recall that this is both supported and contradicted by existing evidence from observational studies of children with CU traits and their caregivers. Recently studies have begun examining mental representations (i.e., cognitive–affective schemas) of the parent–child attachment relationship in children with CU traits. Pasalich et al. [42] reasoned that conduct-problem children with CU traits may be at heightened risk of developing insecure—specifically disorganised and avoidant—patterns of attachment, like their older counterparts. Avoidant attachment is characterised by dismissal of the attachment relationship, and children manifesting this pattern of attachment typically do not seek contact or comfort from their caregiver during times of distress or stress [6]. Unexpectedly, Pasalich et al. [41, 42] did not find a significant association between levels of CU traits and avoidant attachment in their sample of young CU boys; however, the relatively low percentage of avoidant attachment in their sample may have contributed to this finding. Rather, independent of conduct problems, they found a significant link between CU traits and disorganised attachment; wherein children lack any coherent or identifiable strategy to assuage distress [28]. Similarly, a recent study [5] reported a longitudinal relationship between disorganised attachment (age 5) and CU traits (age 7), independent of baseline externalizing behaviour. While it is traditionally thought that parenting that is frightening or frightened, having strong links to unresolved trauma, underpins the development of disorganised attachment-related behaviour [28], child effects are likely. Thus, Pasalich et al. [41] suggested that affective–interpersonal deficits associated with CU traits in children may potentially amplify the effects of atypical or dysfunctional parenting on the development of disorganized attachment.

Finally, emotional responsiveness is related to children’s skills in employing emotional regulatory processes; however the latter has not been studied in the context of emotion recognition paradigms. Children with conduct problems are generally emotionally volatile [47], typically experience high arousal and have poor emotion regulation skills [17]. Relatively little is known however, about the emotion regulation strategies that are used by children with high levels of CU traits. While the range of strategies available to regulate emotions are varied [44], three are commonly used and may be important for this examination of CU traits and emotional responsiveness. Attentional disengagement involves averting the eyes or the head away from the stimulus in order to reduce distress or arousal associated with the stimulus [44]. Given that reduced attention to affective stimuli is one of the core features of psychopathy, and that responses to emotional stimuli in CU children and adults are somewhat normalised when their attention is actively directed to the stimuli, it is possible that disengagement is an emotion regulation strategy that is over-utilised by CU populations. That is, under certain stimulus conditions, children with high CU traits may not pay attention to emotional stimuli as a regulatory strategy, rather than omission based on lack of interest.

A second strategy is to ‘self-soothe’ arousal states by engaging in repetitive grooming type behaviors. These usually involve repetitive manipulation of the mouth, tongue, hands, hair or face, are a well-documented response to emotional arousal and stress in most mammals, and are readily observable in children [7, 26]. In particular, the empathy literature has proposed that self-soothing in response to the plight of another is indicative of self-focused personal distress and comes at the cost of sympathetic responses directed at alleviating the distress of another [20]. Research with infants and toddlers strongly suggests that the incidence of self-soothing behaviours as a strategy for regulation of arousal decreases markedly with age in a typical sample (e.g., [45]), although they have been observed in older children (e.g., [26]). Given the marked decrease in the use of self-soothing behaviours over infancy and toddlerhood, it is possible that the consistent use of self-soothing could indicate some behavioural immaturity, or a difficulty in managing arousal when faced with the intense distress of another.

Third, young children commonly regulate emotion through either verbal or non-verbal communication with their parents [19]. Virtually nothing is known however, about the use of this strategy by children with CP and/or high CU traits. Previous research shows that CP children have poor language competencies for regulating emotion [38], and thus it might be expected that communicative interactions about emotions may be quantitatively or qualitatively lacking for such children. As noted, past research found that children with CP and high CU traits are able and willing to talk about their emotions with caregivers [41, 42]; however, their mothers are more likely to dismiss their emotions [43]. As such, both these regulation strategies, in addition to physical self-soothing behaviours, will be a focus of the current research.

In sum, the literature on emotional responsiveness in children with high CU traits has several inconsistencies and unresolved issues. While lack of emotional responsiveness is a core feature of psychopathy, the extent to which it manifests in children with high CU traits early in development is more controversial. This has significant implications for the design of appropriate clinical interventions for these children. We have reviewed some evidence that indirectly indicates that high CU may be responsive to rich narrative, attachment related cues. This study evaluated whether CP-high-CU children demonstrate shallow affect and poor emotion regulation skills in response to these emotionally complex stimuli. We put the conservative hypothesis that CP-high-CU children would display less affect and emotion regulation than both CP-low-CU and healthy control children in response to both fear and attachment stimuli.

Method

Participants

Ethics approval was from the University of New South Wales HREC 10233 and included full parental informed consent. Participants were N = 76 children recruited from clinical and community sources. N = 56 had been referred to the Child Behaviour Research Clinic, a specialist child mental health clinic in Sydney Australia, for assessment and treatment of behavioural concerns. Of these participants, 74.5 % were male and ages ranged from 4 to 14 years (M = 7.80, SD = 3.11). Assessments were led by clinical psychologists and included a range of parent and teacher measures, and the Diagnostic Interview Schedule for Children, Adolescents, and Parents structured interview (DISCAP; [29, 31]). All met criteria for a DSM-IV disruptive behaviour disorder of Oppositional Defiant Disorder (ODD) or Conduct Disorder (CD) [1]. Secondary diagnoses included Attention Deficit Hyperactivity Disorder (ADHD) (27.3 %), anxiety/depression (7.2 %) and Asperger’s Syndrome (5.7 %). 24 % were stabilised on medication, predominantly stimulants for ADHD. Exclusion criteria for the clinical group included: (a) had no major neurological/physical illness, (b) IQ > 70, and (c) had no formal diagnosis of autism. N = 20 community-based participants were recruited through a private child care service also operating Sydney, Australia; 75 % were male and the ages ranged from 4 to 14 years (M = 7.91, SD = 2.79).

Diagnostic reliability of the clinic-referred sample was assessed by having a team of psychologists and psychiatrists make an independent diagnosis while blind to the primary clinician’s diagnosis. Kappa agreement on primary and secondary diagnoses were .771 and .702 respectively; correlation between severity of ODD/CD was r = .649.

Diagnostic interviews were not available for the community sample however, they were screened to ensure absence of conduct problems and CU traits using mother reports, and using exclusion criteria of (a) current or historical referral or treatment for a behavioural disorder; (b) a score >8 on the UNSW subscale measure of CU traits (see below). These children served as the control-comparison sample.

Report Measures

Child adjustment was measured using parent and teachers reports on the Antisocial Process Screening Device (APSD) [23] and the Strengths and Difficulties Questionnaire (SDQ) [27]. The APSD is a 20-item measure of a child’s antisocial behaviours, with responses given on a three point scale (0 = Not at all true, 1 = Somewhat true, 2 = Definitely true). Sample items include “Blames others for his/her mistakes” and “Lies easily and skilfully”. The SDQ is a 25-item questionnaire rated on a 3 point scale (0 = Not true, 1 = Somewhat true, 2 = Certainly true) regarding the child’s behaviour over the last 6 months or the school year. Subscales include Conduct Problems, Hyperactivity, Emotional Problems, Prosocial Behaviour, and Peer Problems. Sample questions include “Often loses temper” and “Easily distracted, concentration wanders”.

The clinic group was divided into High and Low CU groups using the UNSW system [12] for combining items from parent report on the SDQ and the APSD. This system has a long history showing psychometric validity that produces subscale scores on CU traits, Antisocial Behavior, Hyperactivity, Anxiety/depression, and Peer Problems (e.g., [12]). For the clinic sample, a rating of the child’s CU traits was obtained from each parent. Alpha reliabilities were: mother α = .81, father α = .68, teacher α = .74: agreement between parents was r = .66. Final categorisation was on the basis of mothers’ rating and children who received a score >8 were classified as High CU, which created a cut-off percentile (21 %) corresponding with prevalence estimates of high CU in other conduct-problem samples [24]. Those scoring 8 or below were classified as Low CU. CU traits were measured in the control group using mother reports on the same measures; none of these 20 children were classified as High CU.

Table 1 shows demographic and behavioural adjustment data for the three groups (clinic low CU, clinic high CU, controls) in order to check the validity of their groupings and the presence of any unwanted demographic confounds. The three groups did not differ significantly on demographic variables with the exception of Mother’s education. With regards to measures of overall child adjustment, the groups differed in the expected directions on CU traits, as well as on measures of antisocial behavior, anxiety-depression, hyperactivity and peer problems (see Table 1) and support the validity of the groupings.

Table 1 Percentages, means, standard deviations (SD) and significance of the differences in demographic and child adjustment variables
Full size table

Observation Measures

We used two, 3 min excerpts from Disney’s “The Lion King” which have been shown to elicit high levels of emotional arousal in boys and girls across ages, regardless of prior exposure to the film [4]. In the first clip, the “stampede scene”, the primary emotion is fear, with the film’s protagonist fearing for his life, thus this clip is denoted the “Fear Scene”. It is important to note that within this scene, there is a 2-s interlude in which the villain slaps away a bird that crashes into a rock face and is stunned unconscious. The scene has a slapstick quality that provides a snippet of relief to the ongoing threat, and elicited some amused reactions from some of the children. In the second clip, “Mufasa’s death scene”, the primary emotion is sadness over the loss of an attachment figure, and is therefore called the “Attachment Scene”. For use as comparison ‘neutral’ scene, eight 40-s animations, six pre-existing [8] and two original in the same format, were compiled into a Microsoft PowerPoint presentation that progressed automatically through the slides. The animations feature simple, geometric shapes moving on a plain, white background; some are random movements, others show intentional interactions between the triangles such as dancing and following each other.

Procedure and Coding

All children and parent(s) sat together in front of a laptop (with the screen angled for optimal viewing by the child), with an Aiptek AHD T7 Pro-digital camera elevated behind the laptop screen in full view of the participants. Once the camera was filming, the researcher left the room for the duration of the Lion King and animation excerpts.

Children’s emotion regulation behaviours were coded by two female postgraduate psychology student coders who were blind to the CU traits status of the children. Behaviour was coded using the Child Behaviour Coding Scheme [22]. The target affects were fear (brow raised, eyes widened, lips drawn back), worry-concern (brows lowered, raised cheeks), sadness (inner brow raised, cheeks lowered, lip corners drawn down) and happiness (lip corners raised, cheeks raised). The behaviours of interest were: disengagement, coded when the child’s attention shifted away from the screen and comprised any instance of attentional avoidance using the whole head or shifting attention with the eyes; self-soothing, defined as repetitive manipulation of the mouth, tongue, hands, hair or face, lasting for at least 2 s; and communicative bids, which include the child’s verbal remarks, proximity seeking behaviours and attempts to make eye contact with their parent(s). The manual’s author trained both coders. Each scene, plus the series of animations, was divided into 12 15-s intervals, and target affects and emotion regulation behaviours were coded as present or absent in each interval. Affective and behavioural responses during the video stimuli were coded by two coders for 56 (73 %) children. Inter-rater reliability was high, with a minimum Cronbach’s alpha of .994 (worry-concern during the Attachment Scene) and a maximum of 1.00; that is, there was only a few instances of disagreement throughout.

Results

General Observations of Children’s Behaviour Across Scenes

The children and parents were unanimously highly engaged during the Lion King scenes. Postures were typically leaned forward toward to the screen, talk was minimal, faces were largely serious and concerned, and eye gaze to the screen was rarely broken. Our impressions were strong that the only times children looked away or spoke to parents was in direct response to a high emotional response to the film. That is, we did not observe children being distracted. This is critical in terms of measuring ‘disengagement’ as an emotional regulation strategy. On the contrary, our use of the neutral animation scenes were less successful as they were generally boring to the children and there were high rates of distraction, fidgeting, and talking with parents. We were generally unable to obtain compliance from the children to watch these neutral scenarios, and as such, scoring looking away as a putative emotion regulation strategy (‘disengagement’) is highly problematic in situations where the child is looking away due to boredom. Thus, we did not analyse emotion regulation behaviors for the neutral scenes.

Between Group Differences in Affect

To test for between-group differences in expressed affect for each of the scenes, a 3 (Group) × 3 (Scene) ANCOVA was conducted for each of the affective dependent variables (sadness, worry-concern and happiness). Of the demographic and child adjustment variables, Anxiety-depression was significantly correlated with the target affect of sadness in the Attachment scene (r = −.273; p = .021), so it was used as a covariate in the analyses. Affective expressions of fear occurred too infrequently to be included.

The results are presented in Fig. 1 in which the y-axis is the mean percentage of intervals in which the affect was expressed. There was a significant scene effect for sadness, F(2, 72) = 21.819, p < .005, with significantly more sadness expressed in the Attachment scene than either of the other two. There was also a significant scene effect for worry-concern, F(2, 72) = 9.792, p < .005, with less worry-concern expressed during the Neutral scene. There was also a significant scene effect for happiness, F(2, 72) = 17.38, p < .005, as children expressed more happiness in the Fear scene than the Attachment scene, and more happiness in the Neutral scene than the Fear scene. There was a significant Group × Scene interaction effect for happiness, F(4, 146) = 2.48, p = .046. As can be seen in Fig. 1, High-CU children expressed happiness in a greater percentage of intervals than either the Low-CU or the healthy children during the Fear scene.

Fig. 1
figure 1

Sadness, worry-concern and happiness expressed in each of three scenes by the three groups. Error bars are SE of the mean

Full size image

Between Group Differences in Emotion Regulation Behaviours

Emotion regulation behaviours were analysed with a 3 (Group) × 2 (Scene) ANCOVA, for each of three dependent variables (disengagement, self-soothing and communicative bids). Mother’s education was significantly correlated with the target behaviour of disengagement in both the Fear scene (r = −.554; p < .005) and the Attachment scene (r = −.429, p = .005), and was therefore used as a covariate. Regulation strategies were not coded during the neutral scene as these responses only make sense when observed under levels of high emotional engagement, and not when the child is disinterested, which was more frequent during the Neutral scene across all groups.

Figure 2 depicts the association between group, scene, and regulation strategies. There were two significant main effects for scene: disengagement, F(1, 73) = 65.72, p ≤ .001, and self-soothing, F(1, 73) = 5.89, p = .018, occurred more frequently in the Attachment and Fear scenes respectively. There was no significant difference in communicative bids across the two scenes, F(2, 73) = .264, p = .609. There was a group effect for Disengagement, F(2, 73) = 3.306, p = .042, and a trend for Communicative Bids (p = .06), with the two clinic groups showing higher levels of both these self-regulatory behaviors than the healthy controls.

Fig. 2
figure 2

Disengagement, self-soothing and communicative bids in each of two scenes by the three groups. Error bars are SE of the mean

Full size image

There were no significant interactions between group and scene for disengagement, self-soothing or communicative bids. Figure 2 shows, however, that there were substantially higher levels of both disengagement and communicative bids in the Attachment scene for the clinically-referred children. To test this with greater power, two (post hoc) ANOVAs were conducted comparing the amount of disengagement and communicative bids across groups for the Attachment scene only. Disengagement was significantly higher among the high-CU children compared to the other groups, F(2, 73) = 3.14, p = .049, and communicative bids were significantly lower among the healthy than clinical groups, F(2, 73) = 4.44, p = .015.

Are Group Differences Specific to Level of CU Traits?

We next checked that the differences in emotion regulation were specific to CU traits and not confounded by differing levels of concurrent behavioural and emotional problems that often correlate with level of CU traits. The above analyses were repeated using each of the child adjustment measures of antisocial behaviour, hyperactivity, anxiety-depression, and peer problems as covariates. These did not alter the pattern of findings. Finally, although the difference was not significant, there was a slightly higher percentage of females in the High CU group. Given that it is possible that females may be more expressive of emotions, we felt it important to check this. We were not sufficiently powered to repeat the tests with gender included as a factor, so the analyses were repeated for males and females separately to at least conduct visual inspections of the data. The patterns of results were visually identical for males and females. That is, for both males and females, more happiness was expressed by children with high CU traits in response to the Fear scene, and more communicative bids were made to parents by both clinical groups. Expressions of sadness, worry concern, disengagement and self-soothing were indistinguishable between groups for both genders.

Discussion

This study aimed to investigate whether children with high CU traits are in-fact, callous and unemotional across all situations, or whether they demonstrate emotional responsiveness and emotion regulation strategies in response to complex fear and attachment-related stimuli. We tested the ‘conservative’ hypothesis that children with conduct problems and high CU traits would show less affect and fewer emotion regulation strategies, than conduct problem children with low CU traits and healthy children.

Overall, the results support the rejection of the conservative hypothesis. First, children with CP-high-CU traits expressed more happiness in the Fear scene than the CP-low-CU and healthy children. This is not surprising, however as noted, within the Fear scene there is a 2-s segment that has a slapstick quality that provides a brief relief to fear narrative, and elicited some amused reactions from some of the children. These reactions were more prevalent in the children with high CU traits. There are a number of possible interpretations; first, high CU traits children may have found this scene depicting slapstick violence, inherently funnier. While we could locate no research into CU traits and humour in children, there is evidence that higher psychopathic personality traits in adults are associated with more aggressive humour [50]. Second, it is possible their responsiveness to the humorous interlude was an emotion regulation strategy in which laughing provides a tension relief to the ongoing fear scenes. This seems less likely given that the high CU traits group did not differ from the other groups on the complementary measures of affect and regulation in the fear scene. Third, it is possible that the high CU group were more easily able to disengage from the fear stimuli in order to express humour. Given the extensive evidence that children with high CU traits are less responsive to fear stimuli, it is likely that they were more easily able to express amusement at the slapstick segment despite the overarching fear narrative.

Contrary to this finding for fear, however, children with CP and high CU traits did not differ in their emotional expression during the attachment scene. That is, they showed equal or higher levels of affect than children with CP and low-CU traits and healthy children. This leads us to reject the conservative hypothesis that these children with high CU traits are simply less emotionally responsive across both fear and attachment type scenarios.

A similar pattern was found for emotion regulation behaviours. It was hypothesised that children with CP-high-CU traits would engage in less disengagement and fewer communicative bids during the Attachment scene. On the contrary, there was a trend for the high CU traits children to use disengagement more than the other children. In terms of communicative bids, both CP groups used communicative bids more frequently than the healthy group, thus providing only partial support for this hypothesis.

The greater use of disengagement during the attachment scene among the high CU children is a somewhat controversial finding. As disengagement was coded as any instance of looking away from the stimulus (except when the child looked at the parent during communicative bids), it could be argued that the behaviour is ambiguous. It may reflect an emotion regulation processes used in order to regulate overwhelming emotion as defined by Rothbart et al. [44], or it may indicate a simple lack of interest on the part of the child. While we cannot completely discount the latter interpretation, it is highly unlikely. The current data and previous studies show that these film stimuli are highly engaging [4]. Our clinical sense of watching the children’s responses is that children look away precisely when the intensity of the unfolding emotional scene becomes overwhelming. Thus we conclude that the children with high CU were equally likely to use disengagement to cope with the emotional nature of the attachment scene.

This interpretation is controversial and clearly needs confirmation. There would be several ways of doing this. Children could be assessed for psychophysiological arousal in real time with the observational data to check for positive versus negative correlations between disengagement and arousal [18]. Another useful technique might be to use video-mediated recall strategies [33] in which the child watches themselves watching the video, and are asked to commentate their behaviour in real time with cues such as “I look away here because…”, and so on. Another method for assessing emotional impact is to examine later memory for central versus peripheral features of complex emotional stimuli. Healthy people show a robust tendency to encode central details of emotional events and research is appearing using this technique to assess memory processing associated with high CU traits in children [48].

While awaiting further confirmation of this finding, we are inclined to put forward the hypothesis that a lack of responsiveness to emotional stimuli may be situation specific in children with high CU traits (and perhaps psychopathy more generally). Specificity of responding may be particularly variable across developmental stage and type of stimuli, and that it is not until adolescence and adulthood that a more generalised indifference to the emotions of others becomes the rule. Alternatively, the frequency of contact with attachment figures lessens with age, especially for those high on CU traits, until most interactions are with relative strangers, whose emotions provide little in the way of salience or provocative power.

Some aspects and limitations of this study should be noted. While the clinic children and their caregivers participated at the clinic, the community children were observed at home where there were often more distractors present. Coders were blind to CU traits status but were thus able to discern clinic versus control sample status from the videotapes. While this limits the veracity of differences between healthy and clinic-referred children, it does not impact negatively on the main hypotheses about differences due to levels of CU traits. When determining whether any covariates needed to be included in the model, it was found that both the CP groups had higher anxiety than the No-CP control children. While this finding was unexpected, it has been found by other researchers using relatively young samples [36], and can be interpreted several different ways [17]. There was a large age range in our sample and this may have made any age-specific differences between groups difficult to detect. Our sample was small, especially for the children with conduct problems and high CU traits. This highlights the need for replication especially given recent work showing there may be predictable differences in emotionality within high CU traits groups (e.g., [46]). Finally as noted above, the use of multiple formats to measure the emotional reactions of the children (self-report, memory formation, video-mediated recall, psychophysiological indices) would strengthen the veracity of our conclusions.

The findings of this study extends the findings of Pasalich et al. [39, 41, 42] to show that while their attachment styles may be disrupted, attachment stimuli are highly salient to children with CP and high CU traits. While there is strong evidence that children with CU traits share some characteristics with adult psychopaths [25, 34], the present research adds to the growing evidence that there may be important developmental differences in their emotional and interpersonal functioning. Children with high CU traits may be capable of appropriate emotional responses to complex distress stimuli. The current data indicate that they can find sad attachment scenes distressing, and attempt to regulate this distress by disengaging from the stimulus and, possibly, by seeking comfort from their caregiver.

This suggests that facilitating early life attachment-success may be an important treatment target for children with CP and high CU traits. Given that high CU traits are associated with impaired engagement with emotional ‘core’ stimuli of attachment figures [9, 13, 15], the development of methods for promoting successful attachment or engagement may be challenging. We suggest that the specific stimulus conditions under which these emotional responses occur and fail to occur early in life are critical to etiological models of how psychopathy develops. If these results are replicated, clinical methods by which these emotional responses can be leveraged to support the development of empathic concern for others should be a priority for early intervention research.

Summary

High level of callous–unemotional (CU) traits in children have clinical significance for understanding the causes, prognoses, and treatment needs of children with conduct problems. These traits are defined in part by low emotional responsiveness, especially to the emotions of others. There is controversial evidence, however, that young children with high CU traits do demonstrate emotional responsiveness under certain conditions, namely those associated with attachment threat. We tested this by coding the responses of 4–14 years old clinic-referred children divided into high and low CU traits groups, and children drawn from the community, while they watched film sequences of fearful, attachment-related and neutral stimuli. The high CU traits group were more likely to laugh in response to a brief slapstick interlude to the fearful scene. In the attachment scenario, high CU children expressed similar or trends toward higher emotional responses and emotion regulation strategies, compared to low-CU children and community children.

These results suggest that CU traits in children do not map directly on to the emotional and interpersonal deficits seen in adult psychopathy; rather, they support the idea that high CU children may have the potential for emotional responsiveness to emotional stimuli in attachment contexts, and suggest that facilitating attachment-success is a potential treatment approach for these children. Given that high CU traits are associated with impaired engagement with emotional ‘core’ stimuli of attachment figures, the development of methods for promoting successful attachment or engagement may be challenging. We suggest that the conditions under which these emotional responses occur may be critical to etiological models of how psychopathy develops, and whether these emotional responses can be leveraged to as an early intervention promoting the development of empathic concern for others.