Skin Conductance Responses to a Discrete Threat in Virtual Reality: Associations with Psychopathy and Anxiety

Among children with conduct problems, a subgroup of children who present with a constellation of personality features, referred to as psychopathic traits, show greater risk for aggression and antisocial behavior and other long-term negative outcomes. Most modern conceptualizations of psychopathy specify between two and four dimensions of distinct psychopathic traits, including a narcissism dimension, a callousness dimension, and an impulsivity dimension (Frick & Hare, 2001; Salekin, 2016, 2017). Conceptualizations of psychopathy as early as Cleckley (1976) and Karpman (1941) have described a characteristic absence of fear among those with elevated psychopathic traits, while more recent theoretical accounts have emphasized a so-called ‘low fear’ hypothesis of psychopathy (Lykken, 1957). This characteristic lack of fear may be associated with a lack of concern for the consequences of one’s actions, including the potential for punishment or injury, and may therefore account for the relatively high incidence of aggression and antisocial behavior in these children (Cardinale et al., 2020). However, despite some support in favor of the ‘low fear’ hypothesis of psychopathy, variability in defining and measuring fear, and inconsistent findings on the relationship with distinct psychopathic traits, means that this relationship in children remains somewhat unresolved (Hoppenbrouwers et al., 2016; Sylvers et al., 2011).

Heterogeneity within the psychopathy construct is highlighted by differential associations of the narcissism, callousness, and impulsivity dimensions with an absence of the conscious experience of fear, or reduced autonomic reactivity to threatening or fear-inducing stimuli (Hoppenbrouwers et al., 2016). The method of assessment of fearfulness or fear-related responding in psychopathy research has varied widely, with the majority of studies focusing on self-reported feelings of fear, the ability to recognize and understand fear in others (e.g., fearful facial expression recognition), or the capacity to form learned aversive associations between a neutral and threatening stimulus (Hoppenbrouwers et al., 2016). However, others have relied on the use techniques for assessing reactivity of the autonomic nervous system (ANS) to a threatening or fear inducing stimulus as an index of ‘fearfulness’. Physiological changes that accompany responses to affective stimuli are mediated by the relative actions of the two branches of the ANS: the sympathetic nervous system (SNS), and the parasympathetic nervous system (PNS). Increases in SNS activity are associated with increases in heart rate and greater expenditure of energy, whereas increases in PNS activity are associated with reductions in heart rate and increased conservation of resources. The skin conductance response (SCR) represents an important and reliable indicator of SNS activity. When exposed to stress, a surge of sympathetic activation serves to increase cardiovascular output and skin conductance (Berntson et al., 1991). Importantly, SCRs may vary depending on the nature of the task (Stern et al., 2000), with an increase in SCR indicative of an increase in attention, an increase in emotional arousal (e.g., through an increase in fear, threat, or excitement), or an increase in effortful-control (Critchley, 2002). Although some studies have examined associations between distinct psychopathic traits and SCRs to threatening stimuli in adults, very few studies have examined these relationships in children.

Some of the earliest studies of the relationship between psychopathy and SCR to threat in children focused on differences between children categorized as high versus low psychopathy. For example, children who scored highly for psychopathic traits have been found to show reduced electrodermal responses to threatening stimuli compared to children with low psychopathic traits, while correlational analyses suggested that these reductions were driven by higher factor one (narcissistic/callous-unemotional) and total psychopathy scores, but not factor two (impulsivity) scores (Blair, 1999). Similar results were also reported in another extreme groups design, this time comparing anticipatory SCRs, where it was again found that children in the high psychopathic traits group showed electrodermal hyporeactivity compared to a control group of children with low psychopathic traits (Fung et al., 2005). However, because of differences in antisocial behavior between groups, it remained unclear whether or not these differences were attributable to elevated psychopathic traits, or were more simply a reflection of differences in antisociality more generally.

Several more recent studies have built on these early findings by taking a dimensional approach to understanding the relationship of psychopathic traits with SCR. The majority of these studies have reported reduced SCR in children with higher narcissistic features (variably termed interpersonal or manipulative/deceitful), but not callousness or impulsivity features, during a passive auditory task (Isen et al., 2010), during anticipation of signaled white noise bursts (Wang et al., 2012), or during anticipation and in response to aversive stimuli (Wang et al., 2015). Moreover, across all three studies, these results were only found in boys, but not girls, suggesting that electrodermal hyporeactivity may represent a biological marker of a manipulative and deceitful orientation in males (Isen et al., 2010; Wang et al., 2012, 2015).

However, not all studies have revealed consistent findings. For example, in a sample of 101 young adults, Fanti and colleagues showed that narcissism, but not callousness or impulsivity, was associated with reductions in baseline SCR (Fanti et al., 2017). Yet, all of the trait dimensions were unrelated to SCR reactivity (Fanti et al., 2017), suggesting narcissism is related to a dampened resting arousal that does not generalize to dampened reactivity to threat. Contrasting results have also been reported in a comprehensive study of physiological reactivity in 56 juvenile male offenders, ages 13 to 18 years old, recruited from a juvenile court and detention facility (MacDougall et al., 2019). In this study, narcissism was associated with heightened unsignaled anticipatory SCR, while callousness was associated with smaller unsignaled anticipatory SCR (MacDougall et al., 2019). Neither of these trait dimensions were associated with signaled anticipatory responses, nor signaled or unsignaled reactivity (MacDougall et al., 2019); further, there were no significant relations of daring-impulsive or antisocial psychopathic traits with SCR. Thus, taken together with earlier findings, the pattern of associations between psychopathy dimensions with autonomic reactivity remains relatively unclear.

The studies summarized above varied in terms of assessment of psychopathic traits, with most studies employing measurement tools with a two-factor solution. Although these results are revealing about the relationship of psychopathic with the functioning of the ANS, it is now widely accepted that child psychopathy is composed of three trait dimensions, namely narcissism, callousness, and impulsivity (Frick & Hare, 2001). Further, these studies varied in terms of the measurement of SCR, reporting on anticipatory SCR and event related SCR to a variety of aversive stimuli.

A further dimension that could affect SCRs to threatening stimuli is trait anxiety. Although the relationship of psychopathy with trait anxiety has been shown to vary across different psychopathy trait dimensions (Blonigen et al., 2010), it is notable that psychopathic traits in adolescent males and females co-occur with anxiety at a higher rate than expected (Kubak & Salekin, 2009). Of importance, fear and anxiety are distinguishable at a conceptual and neurobiological level (Sylvers et al., 2011), and it is theoretically plausible that they could show opposing relations with a particular psychopathy trait dimension (Gillespie et al., 2015). Thus, it is important to consider not only the differential relationships of fear and anxiety with psychopathy dimensions, but also whether or not fear and anxiety interact in their relation to psychopathic traits.

Table 1 shows the descriptive statistics. SCR to the drop was positively skewed (Z = 5.13). After performing a square root transformation, skewness was reduced to 0.595 (SE = 0.281). Examination of violin plots and boxplots revealed that the variable approximated a normal distribution with no outliers. Means and standard deviations are shown in Table 1 and SCR mean and standard deviation are given for the non-transformed variable to aid comparison with other studies.

Correlations noted in Table 1 showed that, although not significant, greater SCR to the drop was associated with higher levels of narcissism and lower levels of callousness. Narcissism was positively and significantly associated with child-reported anxiety, but the positive association between impulsivity and anxiety did not reach significance; nor did the negative association between callousness and anxiety. There was a small positive association between SCR and anxiety, but it was not significant. Of importance, there were no significant associations between SCR and reports of enjoyment of the roller coaster experience (r = 0.18, p = 0.156), VR matching reality (r = 0.19, p = 0.140), or absorption in the VR experience (r = 0.03, p = 0.817). This suggests that the SCR to the drop was not influenced by positive affect or any variation in participants’ perceptions of the roller coaster as matching reality. Teacher reports indicated that boys showed higher levels of callousness, but because of the low number of girls recruited to this study we did not control for gender.

Psychopathy is a multidimensional construct consisting of narcissistic, callous, and impulsive tendencies (Salekin, 2016, 2017). These trait dimensions have been found to show differential relations with indices of emotion processing and physiological reactivity, with a particular emphasis on the experience of fear. In the present study, we aimed to examine whether fear and anxiety act in concert or reciprocally to explain variance in narcissism, callousness, and impulsive psychopathic traits in adolescence. We used a measure of physiological reactivity to a VR roller coaster drop event as an index of fear reactivity, and asked participants to self-report their experiences of anxiety. In line with expectations, we found that callousness was related to a double-dose of low SNS reactivity, in the form of low SCR to a fear-inducing drop on a VR roller coaster, and low anxiety. Although zero-order associations showed that narcissism was positively associated with anxiety, there were no significant relations of either narcissism or impulsivity with SCR. The interaction effects between anxiety and SCR to the drop in predicting narcissism and impulsivity were also non-significant. Although narcissism and impulsivity showed nonsignificant positive associations with anxiety, these associations were not moderated by SCR. These findings further our understanding of psychopathy dimensions and affective responses to threat, both in terms of peripheral psychophysiological responding and in terms of trait-anxiety.

Prior to interpreting the findings fully, it is necessary to outline the mechanisms of SCR. SCR has diverse mechanisms as an index of SNS reactivity. The VR task in the present study involved sitting on a VR roller coaster, watching the roller coaster in first-person perspective while it ascended slowly to the top of a mountain, and then watching it descend quickly through a fantasy valley. An increase in SC (i.e., SCR) could indicate an increase in attention, an increase in emotional arousal (through an increase in fear, threat, or excitement), or an increase in effortful control that might lead to an increase in peripheral SNS activation. Since we marked the timing of the event across all participants, we ensured that physiological responses (i.e., the SCR) were locked to the drop event. However, the emotional salience of the event can also vary based on novelty, familiarity, personal significance, potential anticipation of reward, memory recall, and attentional and cognitive effort (Critchley, 2002). The results that we report here suggest that the associations we found are unlikely to reflect enjoyment or absorption with the VR roller coaster, as there were no significant associations between SCR to the drop and enjoyment and absorption with the stimulus.

SCR was chosen for its role in measuring fearful arousal. Prior research has shown that low event-related SCR was associated with greater ventromedial prefrontal cortex (vmPFC) activation (Zhang et al., 2014), suggesting that vmPFC activation helps to curb fear-related responses (Nili et al., 2010). Additionally, while SCR appears to be negatively associated with activity largely in the posterior vmPFC, positive associations with SCR were mainly found in areas of the brain associated with anxiety (Zhang et al., 2014). Based on the above, we expect that SCR to the roller coaster drop reflects a measure of threat or fearful arousal.

Our results for the narcissism dimension are consistent with earlier work showing that participants who score higher on the narcissism dimension experience greater anxiety (Barry & Malkin, 2010; Lee-Rowland et al., 2020). Lee-Rowland et al. (2020) proposed that youths with high levels of narcissism may exert superiority over others and protect their self-image due to a sense of insecurity and high levels of anxiety. This tendency to exert superiority over others may lead to increased aggressiveness and attempts to dominate others (Barry et al., 2007; Golmaryami & Barry, 2010; Washburn et al., 2004). The absence of an association between narcissism and SCR is surprising in light of previous research showing a reduction in SCR in anticipation of, and in response to, an aversive stimulus (Isen et al., 2010; Wang et al., 2012, 2015). Recently, for example, MacDougall et al. (2019) showed that grandiose/manipulative traits (roughly equivalent to psychopathy-linked narcissism) were associated with heightened SCR during anticipation of, and in response to, an aversive event (white noise). However, previous studies have predominantly tested these effects in samples of boys, and our findings suggest that these effects may differ in mixed samples of boys and girls. It is also noteworthy that the task used in the present study could be interpreted as thrilling rather than aversive, which may elicit a different pattern of relations with physiological responding.

In contrast to the results for narcissism, callousness was related to measures of emotional experience that indicate low peripheral reactivity in response to a fear inducing stimulus as well as low anxiety. Our results are largely consistent with the findings of MacDougall et al. (2019). In their study, callousness was related to lower SCR during the anticipatory period where youths were not alerted to the subsequent arrival of the aversive stimulus (MacDougall et al., 2019). Yet, our own results go beyond these findings to also show that anxiety was associated with callousness, but only at low levels of SCR to the fear inducing event. Thus, it is possible that youths who are more callous might down-regulate their arousal in the face of fear or threat and could be less prone to feelings of anxiety. Indeed, Thomson et al. (2019) found evidence for heightened emotion regulation for youths high on callous-unemotional traits when measuring PNS activity (i.e., augmentation of respiratory sinus arrhythmia) while experiencing the entire duration of a real-3D TV and VR roller coaster. This has led to proposals regarding the usefulness of measuring ANS activity or reactivity to aid the assessment of callousness and interventions targeting socioemotional functioning during social affiliation (Wagner & Waller, 2020). Our findings reported here may help to explain a link between callousness and insensitivity to punishment (Muñoz & Frick, 2012), whereby youths who fail to respond to fear or threat with a sympathetic response may find it hard to remember and learn from negative events, leading to difficulties in early socialization (Fowles & Kochanska, 2000; Kochanska, 1993).

Our results showed that the impulsivity dimension was unrelated to either SCR to the fear inducing event or anxiety. These findings are largely consistent with earlier work, with most reports documenting the absence of statistically significant association between SCR and the impulsivity features of psychopathy in children (Isen et al., 2010; MacDougall et al., 2019). Taken together, these findings suggest that electrodermal reactivity may not represent a reliable, biological marker of impulsivity. The absence of an association between impulsivity and SCR in study might also reflect the nature of the stimulus used in this study, which employed a thrilling roller coaster drop, and earlier work that has presented anger inducing stimuli and has required a behavioral response (Tangney et al., 2004).

There are several limitations to note in interpreting our findings. First, although this was a sample with emotional and behavioral difficulties, we did not perform any diagnostic testing for conduct problems. It is important to note that research suggests that findings differ for those with callous-unemotional traits in the presence versus absence of conduct disorder (Fanti et al., 2017, 2018), and diagnoses of conduct disorder should therefore be considered in future research. Our sample also showed high rates of ADHD, and some participants had existing diagnoses of ASD, reported taking medication, and had histories of abuse. Although rates of comorbid disorders and histories of medication and abuse are common in samples of children with emotional and behavioral difficulties, and our sample is therefore representative of the realities of working with these populations, these factors may make replication in other samples more difficult. Other problems in affective functioning, for example low or depressed mood, were not assessed here but should be examined in future studies.

Other limitations related to the measures that we used. For example, our measure of callousness was only moderately internally consistent. Despite some doubts in the present sample, this subscale has shown stability over time, as would be expected for a measure of personality traits (Muñoz & Frick, 2007), and shows prospective relations with antisocial behavior over and above the other psychopathy trait dimensions (Muñoz & Frick, 2007). A further limitation arises from the use of a single drop event in the VR based roller coaster, meaning that our analyses were based on a single-item measure of SCR. Although data from other drop events during the roller coaster could be used, problems with habituation led to little variance in reactivity to the drop. This task was also used to capture reactivity to a fear inducing event, and the presence versus absence of fear has been linked in particular with the callousness features of psychopathy (Cardinale et al., 2020). Thus, other VR settings that elicit emotions other than fear or the thrill and excitement of a roller coaster may be associated with different patterns of physiological reactivity, and different relations with psychopathic traits. Finally, we had a relatively small sample owing to the difficulty in conducting psychophysiological studies using VR within schools. However, one of the strengths of the study is the relationships with school headteachers and teachers. The cooperation with multiple school officials also meant that recruitment of a specialist group of youths was successful.

The VR context provided us with an opportunity to examine children’s psychophysiological responses under more realistic conditions compared to using film clips, static pictures, or contrived threatening events (where there are no real consequences). It would have been difficult to collect psychophysiological responses on a real-life roller coaster, but a fantasy world can easily be created that leads to physiological sensations in VR. One noteworthy prior study has also examined responses to a real-life fear-inducing event. During five emotionally evocative life events (e.g., getting inoculated), adolescent youths with psychopathic traits showed specific reductions in self-reported fearfulness (Marsh et al., 2011). However, in that study, the authors did not examine the consistency of self-reported fear with physiological reactivity. The benefits of using VR have been argued to be “making the impossible, possible” (to create environments that would be hard or impossible to recreate outside of the realm of imagination), “making the possible, safer” (one can explore dangerous or threatening scenarios in complete safety) and “making the possible more safe, more practical and more effective” (Negen, 2018). Thus, using VR could have intervention and research implications, especially when used in combination with psychophysiological techniques to understand arousal and affect regulation processes under more life-like conditions, as has been attempted for borderline personality traits (McLachlan et al., 2021).

Youths who score highly on different psychopathy trait dimensions may present cause for concern in different ways. By taking a dimensional approach to understanding the relationship of narcissistic, callous, and impulsive psychopathic traits with SCR, we were able to demonstrate differential relations of psychopathic traits with fear reactivity and self-reported anxiety. In sum, we found that narcissism, in this multi-health problem group, was associated with anxiety but not with the response to a thrilling event of a roller coaster, while callousness was related to a dampening of emotions across measures of fear reactivity and anxiety, and impulsivity was not significantly associated with either SCR or anxiety. The finding that higher levels of the affective component of psychopathy (i.e., callousness) were related to low anxiety and low SCR may be of importance for understanding the aggressive and antisocial behaviors associated with callousness, but the uniqueness of our sample means that more research is needed. As noted by Fanti (2018), the psychopathy trait dimensions are differentially associated with emotional and physiological reactivity, and our findings suggest that simultaneous measurement of internal emotional states may also be of considerable importance. Future work should also consider that ANS activation been found to be inconsistently related to brain activation across peripheral organs (Eisenbarth et al., 2016), and so triangulating across different peripheral ANS measures (i.e., SCR, respiratory sinus arrhythmia, etc.) may be fruitful. Using VR may be one way to not only test emotional responses to frightening events, but it may also be of potential use in interventions.