Introduction
Children with conduct problems (CP) display a range of antisocial behaviours including bullying and manipulation, physical aggression, and violation of societal rules and norms (Frick
2016). Children with CP have a greater risk of physical and mental health problems, difficulties with personal relationships, as well as reduced employment and increased criminality in adulthood (Frick
2016; Rivenbark et al.
2018; Wertz et al.
2018). They require more support from specialist education provisions, have increased use of health and social care services, and increased contact with the criminal justice system which creates a significant financial burden for society (D’Amico et al.
2014; Frick
2016; Scott et al.
2001). This has created an impetus for earlier and more targeted intervention strategies to halt the development of CP for the good of the individual and society (Rivenbark et al.
2018; Stellwagen and Kerig
2013).
Considerable research has demonstrated that children with CP are a heterogeneous group and one way of understanding the heterogeneity of CP behaviours is to consider the role of callous-unemotional (CU) traits (Frick et al.
2014; Frick and Viding
2009; Viding and McCrory
2015). Children with CP and high levels of CU (CP/HCU) display a callous lack of remorse and guilt and marked deficits in empathy (Frick et al.
2014; Viding and McCrory
2015) and are thought to be at an increased risk of developing psychopathy in adulthood (Frick et al.
2014; Frick and Viding
2009). Children with CP/HCU not only display impulsive and reactive antisocial actions, but also commit calculated acts of aggression with little regard for other people’s feelings (Frick et al.
2014; Blair et al.
2014; Pardini and Byrd
2012). In contrast, children with CP and low levels of CU (CP/LCU) do not have pronounced deficits in empathy and remorse and often commit acts of aggression that have clear environmental triggers, such as perceived threat or frustration (Frick and Viding
2009; Blair et al.
2014). Measurement of CU traits (termed ‘Limited Prosocial Emotions’) was included in the latest edition of the Diagnostic and Statistical Manual of Mental Disorders 5th Edition (DSM-5) as a specifier for children with Conduct Disorder (American Psychiatric Association
2013).
A substantial amount of work has focussed on how children with CP/HCU process emotional signals from others. Children with CP/HCU have been found to have difficulties in processing emotional information, such as, having reduced neural responses when viewing other people in pain (Lockwood et al.
2013), reduced neural and behavioural responses to laughter (O’Nions et al.
2017), and difficulties in responding to and resonating with other people’s fear and sadness (Blair et al.
2014; Frick et al.
2014; Lozier et al.
2014; Viding et al.
2012). These difficulties, particularly difficulties with resonating with other people’s emotions, might in part explain why children with CP/HCU are able to engage in acts of aggression and violence and why they do not form typical affiliative relationships (Blair et al.
2014; Viding and McCrory
2019).
Another important aspect of social and emotional processing involves mentalising, which is the ability to understand the thoughts, intentions and feelings of other people (Fonagy and Allison
2012; Frith and Frith
2006). Mentalising is essential for all aspects of social interactions, allowing one to consider not only one’s own perspective, but also the various perspectives of others (Choudhury et al.
2006). Several studies have reported that children with CP/HCU are able to make accurate mental state inferences when the mentalising task does not require the participants to consider affective content (Anastassiou-Hadjicharalambous and Warden
2008; Jones et al.
2010; O’Nions et al.
2014; Schwenck et al.
2012). For example, Jones et al. (
2010) found that CP/HCU children have difficulties with affective resonance, but not with cognitive perspective taking (i.e. mentalising without affective content), with the opposite pattern reported for children on the autism spectrum. Children with CP/LCU did not differ from TD peers on either affective resonance or cognitive perspective taking in this study. Other studies have reported similarly spared ability in making mental state inferences when children with CP/HCU are not required to mentalise about emotions (Anastassiou-Hadjicharalambous and Warden
2008; Schwenck et al.
2012). Additionally, O’Nions et al. (
2014) reported that children with CP/HCU show recruitment of similar brain regions to TD peers when required to process scenarios that require mentalising, but which do not have affective content, whereas children on the autism spectrum show reduced activity in brain regions associated with mentalising compared with TD peers. This pattern of findings makes sense in the light of what is known about the behaviour of children with CP/HCU. They are able to successfully manipulate others for personal gain, which would not be possible without the ability to mentalise, however, they display clear deficits resonating with others’ feelings.
Although the basic ability to mentalise has been found to be intact, behaviours of children with CP/HCU suggest that they have a reduced propensity to mentalise (Viding and McCrory
2019). They tend to be more self-focused and can aggress even when someone is showing distress, especially if they stand to gain something (Jones et al.
2010; Pardini et al.
2003). A recent study by Drayton et al. (
2018) has found that adult psychopaths can deliberately take the perspective of others, which may help them to manipulate others, but do not always spontaneously do so. Drayton et al. (
2018) proposed that this pattern of functioning may enable individuals with psychopathy to avoid processing the emotional consequences of their antisocial behaviour towards other people or even orienting to other people’s needs in the first place. It seems that individuals with psychopathy can take on the perspective of others when it helps them achieve a goal but ignore it when it is not useful to them. In other words, part of the reason why individuals with psychopathy (or at risk of developing psychopathy) may so readily be able to prioritise ‘looking after number one’ could be due to their reduced tendency to consider other minds and/or make mental state inferences, while having the cognitive machinery to do so when it serves their own needs (Drayton et al.
2018).
The aim of the current study was to assess mentalising using three different tasks. We administered the Movie Assessment of Social Cognition (MASC) which asks participants to assess characters’ mental states after watching them interacting in a video (Dziobek et al.
2006). This task was selected as it presents a variety of information and cues (social, verbal, physical) and participants are asked to make assessments of thoughts, feelings, and intentions in ‘real-time’, similar to what one might encounter in real-life interactions with others (Sharp et al.
2011). It is thought to assess the ability/propensity to incorporate judgements about the protagonists’ minds into inferences about their mental states (e.g., whether one updates one’s estimate of the likelihood a character will be suspicious of another character based on whether the former character is thought to be paranoid or easy-going; Conway et al.
2019a,
b). The MASC has previously been administered to a small sample of children with behaviour problems in a mainstream school (Körner et al.
2009). In this study, behaviour problems were associated with a reduced number of correct mentalising responses, however the rating of behaviour problems was limited to teacher reports on a single measure and no quantification of CU traits was provided (Körner et al.
2009). We also asked children to complete a standardised mind-mindedness task, which assesses the tendency to think about the minds of peers that are relevant to the participant. This task requires the participants to spontaneously describe a good friend, with no restrictions or limitations on their description. Previous research has found that young adults are more likely to make mind-related comments about someone they know personally, rather than a stranger, with intimacy providing greater knowledge of and ease of access to the person’s mental states (Meins et al.
2014). Children with CP/HCU may not be motivated to mentalise about strangers unless they can personally gain something out of it, however it may be less effortful and more instrumentally useful for them to consider the minds of peers that they regularly interact with. Finally, we administered the Social Judgement Task (SJT), an illustrated mentalising task that asked the participants to report what other children would think about them, if they engaged in a negative interaction with a fictional peer. The negative interaction scenarios in the SJT were developed to assess whether acting antisocially may be, in part, explained by difficulty in accurately predicting how the antisocial acts are viewed by others. This task provides insight into whether children with CP/HCU can infer what other people, specifically peers, think when they engage in social transgressions against others. The participants are also asked to report on the likelihood of committing acts described in the scenarios, providing a possible index of acting antisocially, despite knowing how it is viewed by others.
We chose to focus on groups of boys with CP/HCU and CP/LCU instead of conducting continuous analyses for the following reasons: 1) Effects of having distinct subgroups of children with CP as divided on CU traits do not often emerge as interactions and can instead lead to suppressor effects in correlational analyses (Frick
2012); 2) We know that bivariate normality does not apply to CP and CU distributions where high CU traits almost invariably denote high levels of CP, but not the other way around (Fontaine et al.
2011). Dichotomizing leads to reduction of power in the case of bivariate normality (Cohen
1983), but we know that bivariate normality does not apply to CU traits and CP; 3) The median split approach has, in the past, successfully delineated groups of children with CP who have different cognitive-affective processing patterns – often in a manner that would lead to the two groups cancelling each other out if pooled into a single CP group for comparison with typically developing children, or which do not necessarily emerge in dimensional analysis in community samples that represent the whole spectrum of scores. The child/group centric analyses also make it easier to interpret the translational relevance of findings, which is more challenging when examining potential suppressor effects. Some previous cross-sectional research has found higher mean levels of CU traits in older adolescents (Essau et al.
2006), although this is not evident in longitudinal data (Pardini and Loeber
2008). However, to ensure that age differences were not accounting for the findings, the groups were matched on age, with comparable representation across the age bands of the sample.
Although experimental findings indicate that individuals with or at risk of developing psychopathy have an intact ability to represent other minds (Anastassiou-Hadjicharalambous and Warden
2008; Jones et al.
2010; O’Nions et al.
2014; Schwenck et al.
2012), their behaviour suggests a reduced propensity to consider others. At the cognitive level this may manifest as: i) a reduced ability/propensity to incorporate mind type into mental state inference (as assessed by the MASC); ii) a reduced propensity to represent the minds of others (as indexed by the Mind Mindedness task); or iii) reduced ability to infer what other people think about them when they engage in social transgressions (as assessed by the SJT). The purpose of this study was to investigate each of these possibilities.
Discussion
Boys with CP/HCU had difficulty mentalising (as compared with TD and CP/LCU boys) when they performed a complex, ecologically valid task which indexed the ability/propensity to incorporate judgements about another’s mind type into inferences about their mental state (the MASC task). However, they did not differ from TD boys in their propensity to represent the minds of their friends when asked to describe them, or in their ability to understand that other children would think negatively about someone committing antisocial acts. Boys with CP/LCU did not differ from TD boys on performance in any of the three tasks. These findings provide a more nuanced picture of mentalising in boys with CP. Overall, they are in line with prior studies suggesting an intact ability to mentalise in children with CP, including those with CP/HCU, especially if there is no requirement to consider other people’s feelings. These findings also suggest that despite having the ability, boys with CP/HCU may have a reduced propensity to mentalise than their peers. They may only deploy this ability spontaneously if it does not require them to process complex information or if it is of instrumental benefit to themselves.
In line with our hypotheses, boys with CP/HCU had difficulty with the MASC task, in particular with the ‘intentions’ questions (assessing cognitive mentalising). MASC, unlike most assessments of mentalising, depicts people interacting in real life situations. Task performance depends on the ability/propensity to incorporate information about each character’s mind in order to make accurate mental state inferences during an observed ‘live’ interaction (Conway et al.
2019b; Dziobek et al.
2006). The effect of group on the ‘intentions’ questions was no longer significant after adjusting for cognitive empathy (as measured by the BES cognitive scale). Although the groups only showed a trend level difference on BES cognitive empathy, the CP/HCU boys had the lowest scores on this measure and the BES cognitive empathy scale taps into ability/propensity to incorporate information about other people’s minds to make accurate mental state inferences. It therefore follows that cognitive empathy would be having effect on correct responding to ‘intentions’ or cognitive items in the MASC as both are focussed on understanding the perspective of others. CP/HCU children may not be interested in others’ minds unless other people are instrumentally valuable, or they have a mind that is vulnerable or easy to manipulate. It could also be that the characteristics of children with CP/HCU mean that they will experience a restricted range of social interactions with other people, which may in turn reduce the number of types of mind to which CP/HCU children are exposed. While CP/HCU boys had clear difficulties with the ‘intentions’ questions in the MASC, they did not significantly differ from TD or CP/LCU participants in spontaneously mentalising about ‘feelings’ (affective mentalising). Although this may seem surprising, it is important to note that all groups had difficulties with the ‘feelings’ questions and it is likely that no group differences emerged because of a floor effect. It would, therefore, be inappropriate to conclude that boys with CP/HCU do well in spontaneously mentalising about feelings (in fact their rate of mentalising about feelings was very similar to their rate of mentalising about intentions). Instead, it appears that adolescent boys from similar SES backgrounds and of similar cognitive ability all show low levels of spontaneous mentalising about emotions.
Boys with CP/HCU showed reduced spontaneous mentalising about the interactions of strangers in the MASC task, but there were no group differences when boys were asked to spontaneously mentalise about a friend. CP/HCU boys appear similar to CP/LCU and TD peers in their propensity to represent friends’ minds. This may be explained by the greater knowledge one has about friends rather than someone with whom there is no personal relationship (Meins et al.
2014). Familiarity makes it easier to represent the mental states of friends. CP/HCU boys may also be more motivated to represent the minds of friends, as understanding friends’ point of view could be instrumentally valuable, if for no other reason than for successful manipulation. It may also be that CP/HCU have a similar mind type to their friends which makes it easier to infer mental states (Conway et al.
2019b). There were no group differences on affective mind-minded comments, but as was found with the MASC where all groups had difficulty with the ‘feelings’ questions, all groups had low levels of mentalising about their friends’ feelings and emotions. It would, therefore, be inaccurate to conclude that CP/HCU boys are inclined to consider their friends feelings when describing them.
Although the CP/HCU group had difficulty with the MASC task, they had an intact ability to infer the thoughts of others regarding engagement in antisocial actions. Boys with CP/HCU knew just as well as typically developing boys that peers would find antisocial acts unacceptable. This indicates that they can understand what is wrong and more critically how that is perceived by their peers. The SJT task does not require any inference of others’ feelings and it may be helpful for future research studies to include an affective component to explore whether group differences occur when children are asked how they might feel if they acted as the antisocial story described or how peers would feel about them if they acted antisocially. Interestingly, CP/HCU boys were not more likely to say they would act antisocially, as described in the story, than their TD or CP/LCU peers. It is instrumentally valuable to consider the thoughts of others with regard to antisocial actions and only execute such actions when the outcome is judged to be sufficiently valuable to discard the displeasure of others. In this case, it may not have been worth discarding the potential displeasure of the researcher given that there was nothing tangible to be gained by reporting that they would be likely to act as the story described. It is not adaptive to act in an antisocial way at all times, as this is likely to preclude taking maximal advantage of someone.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.