In this study, youths with psychopathic traits showed reduced amygdala responsiveness to legal actions relative to healthy youths. Moreover, they showed reduced amygdala-orbitofrontal cortex connectivity relative to healthy youths during task performance. They did not, however, differ from healthy youths in terms of dorsomedial or lateral frontal cortex activation during the task.
It has been argued that moral judgments reflect the amygdala’s role in stimulus-reinforcement learning. The amygdala’s role is thought to involve enabling the individual to learn the goodness or badness of actions and feeding forward reinforcement expectancy information to orbitofrontal cortex to facilitate decision-making (Blair, 2007
). These systems are considered dysfunctional in individuals with psychopathic traits, as studies with youths with psychopathic traits (Finger et al., 2008
; Marsh et al., 2008
; Jones et al., 2009
) and adults with psychopathic traits (Kiehl et al., 2001
; Gordon et al., 2004
; Birbaumer et al., 2005
) have demonstrated. Recently, adults with high psychopathy scores have been reported to show reduced amygdala activity during moral decision making (Glenn et al., 2009
) and to show reduced correspondence between amygdala activation and judgments of moral severity (Harenski et al., 2010
). The current data extend these findings by demonstrating that the region of the amygdala showing reduced responding to legal actions in youths with psychopathic traits showed, as predicted, reduced connectivity with orbitofrontal cortex. The extension of these findings to a youth sample is important because it indicates that the identified pathophysiology is unlikely to reflect a secondary consequence of the disorder (e.g., the increased drug abuse that is prevalent in adult samples) but rather a developmental feature of the pathology.
It should be noted that the moral decision making paradigm used in the study of adults with high psychopathy scores (Glenn et al., 2009
) is a more complex task than the moral IAT task used here. The study by Glenn et al (2009)
used variants of what have been termed “trolley problems” (cf. Greene et al., 2001
) that force decisions between competing moral objectives. In contrast, the current task involved the participants making simple judgments regarding the legality of items. Importantly, though, both types of task may be reliant on a form of an emotional “automatic moral attitude”.
At the neural level, this automatic moral attitude is thought to reflect the amygdala’s response to the conditioned stimulus that is the individual’s representation of the moral action (whether prosocial or antisocial) and the representation of this valence information within orbitofrontal cortex (Blair, 2007
). In the context of “trolley problems”, it is argued that this emotional automatic moral attitude is the basis of the distinction between personal dilemmas and impersonal dilemmas, which are distinguished by the salience of the harm to a victim. In the current task, negative and positive emotional automatic moral attitudes may have been activated by reading the antisocial and prosocial actions, respectively.
The claim is not that youths with psychopathic tendencies are selectively impaired in representing the value of legal items. Rather, youths with psychopathic traits are thought to face difficulties associating the appropriate valence with objects and actions generally (Blair, 2007
). In the current study, this manifested as a pervasive pattern of atypical activation in response to positively valenced items (legal actions and positive objects). Other work has also identified aberrant neural responses to positively valenced cues in youths with psychopathic tendencies (Finger et al., 2011
) or youth with conduct disorder (psychopathy unspecified; Rubia et al., 2009b
; Crowley et al., 2010
). Currently, it is unclear why the present study, and these previous works, identified particular problems with positively valenced items whereas other work has identified difficulties with negatively valenced items (e.g., Birbaumer et al., 2005
; Finger et al., 2008
; Marsh et al., 2008
; Passamonti et al., 2010
). These differences may reflect the computational specifics of the individual tasks used and is a current focus of research in our group.
Three features of the results are worth considering further. First is the apparent selectivity of the result: youths with psychopathic traits showed reduced amygdala responsiveness relative to comparison youths to the legal items but not to the illegal items. We anticipated a significant group difference for illegal items also. The absence of a group difference for illegal actions may reflect a type II error. Previous behavioral work demonstrates significant impairment in the processing of antisocial actions in individuals with psychopathic traits (Blair, 1995
; Blair et al., 2001
). And Glenn and colleagues (2009)
found that psychopathy was associated with reduced amygdala activation in response to actions associated with salient harm.
Second, previous IAT imaging studies typically find greater BOLD responses in dorsomedial frontal cortex (including dorsal anterior cingulate cortex) and lateral frontal cortex during incongruent than congruent trials (Chee et al., 2000
; Luo et al., 2006
; Beer et al., 2008
). Dorsomedial frontal cortex is implicated in mediating response conflict (Cohen et al., 2000
; Kerns et al., 2004
), potentially by recruiting lateral frontal cortex to augment the representation of relevant stimulus features within temporal cortex (MacDonald et al., 2000
; Garavan et al., 2002
). It has been argued increased dorsomedial frontal and dorsal anterior cingulate cortex activity during the IAT effect reflects increased response conflict during incongruent trials (Luo et al., 2006
). In the current study, there was a significant main effect of congruence within dorsomedial frontal cortex that extended into dorsal anterior cingulate cortex and lateral frontal cortex, but no group-by-congruence interaction. This suggests a lack of deficits in anterior cingulate cortex in youths with psychopathic traits during tasks of this type. This finding corroborates similar findings in prior imaging studies in these youths (Finger et al., 2008
Third, and in line with predictions, healthy comparison youths showed positive connectivity between amygdala and orbitofrontal cortex. This may reflect the amygdala’s role in feeding forward reinforcement expectancy information to orbitofrontal cortex to guide decision-making (Blair, 2007
). Youths with psychopathic traits showed significantly less positive functional connectivity between these two regions. In fact, in contrast to predictions, the youths with psychopathic traits showed significant negative connectivity between the amygdala and orbitofrontal cortex. This suggests a relationship between amygdala and orbitofrontal cortex activity that is inverse to that of healthy comparison individuals. This may reflect inappropriate interactions between these two regions in youths with psychopathic traits. Alternatively, this might reflect atypical regulatory activity over the amygdala by orbitofrontal cortex in youths with psychopathic traits (cf. Urry et al., 2006
). Although we believe this alternative hypothesis is premature, further research will be conducted to examine emotion regulation in youths with psychopathic traits.
Several additional concerns should be noted regarding the current results. First, no behavioral groups differences were observed. This is consistent with the results of prior studies that have measured simple permissibility judgments like those assessed here. One prior behavioral study used a moral-judgments IAT similar to our own, and the authors reported group differences in psychopathic murderers, but not in psychopaths who were not murderers (Gray et al., 2003
). In a previous fMRI study of moral reasoning in psychopaths, performance differences were not seen, although higher psychopathic tendencies were associated with reduced amygdala activity (Glenn et al., 2009
). By contrast, previous studies testing somewhat more sophisticated moral reasoning have found performance differences in individuals with increased psychopathic traits (Blair, 1995
; Blair et al., 2001
). We interpret group differences in neural activation patterns in the absence of behavioral differences as potentially reflecting differences in strategies used by psychopathic and non-psychopathic adolescents during moral judgments. Perhaps some participants relied on semantic knowledge rather than affect-based assessments to perform the task. If prevalent, the use of such a strategy might also help to account for the lack of between-group differences in amygdala activation during illegal-word trials.
Second, one aspect of the functional connectivity analysis approach used here instead of, for example, a psychophysiological interaction (PPI) analysis, is that it does not allow us to distinguish group differences in connectivity related to the performance of the IAT task from group differences in baseline connectivity. However, a PPI analysis would have been less appropriate given our task design. According to our theoretical position, all conditions relied on amygdala-orbitofrontal cortex interaction. Although group differences in amygdala activation may be greatest for legal items, we did not predict that group differences in amygdala-orbitofrontal cortex connectivity would vary across conditions. Thus, our analysis, in which functional connectivity was examined across all conditions of the IAT task, was able to test our hypotheses. Future work might, however, include a low-level control condition. This would allow for better examination of whether reduced amygdala-orbitofrontal cortex connectivity reflects group differences in connectivity specific to the task or to particular trials or group differences in baseline connectivity.
Third, the current study contrasted youths with disruptive behavior disorders and psychopathic traits with healthy comparison youths—it did not include a comparison group of youths with disruptive behavior disorders with low levels of psychopathic traits. We assume that the atypical amygdala responsiveness and amygdala-orbitofrontal cortex connectivity observed here is causally related to the emergence of psychopathic traits (Blair, 2007
). However, without this second comparison group, we cannot be certain whether the observed pathophysiology relates to youths with disruptive behavior disorders generally or only those with disruptive behavior disorders and psychopathic traits. It will be important for future investigations to compare patterns of neural activation across subgroups of children with conduct problems, given the heterogeneity observed in previous studies of this population (Frick and White, 2008
; Passamonti et al., 2010
). Moreover, it will be important to determine whether the present results extend to psychopathic adults, who are typically selected using a standard PCL-R cutoff of 30/40 points (higher than the 20/40 points cutoff we employed).
It will be recalled that 6 children with psychopathic traits in this study were taking psychotropic medications. To rule out the possibility that these medications were the cause of the patterns we observed, we repeated our primary ANOVA excluding these children and observed nearly identical Group × Legality effects in the amygdala as had been identified in the original analysis, (xyz = 17, -1, -22, 4 voxels), F(1,26)=9.92, P<0.005. This suggests that group differences we observed were not the result of group differences in medication status. The fact that similar results were obtained after nearly halving our group of adolescents with psychopathic traits supports the stability of the patterns of activation we have identified. However, it should be noted that small sample sizes such as that used in the present study increase the instability of data, thus increasing the risk of identifying patterns of activation that result from chance.
Fourth, disruptive behavior disorders (CD and ODD) are frequently comorbid with Attention Deficit Hyperactivity Disorder (ADHD) (Taylor et al., 1986
). In this study, 64% of youths with psychopathic traits also presented with ADHD. It could therefore be argued that group differences reflect ADHD rather than psychopathic traits. Because of this concern, our first phase of studies included two comparison groups of participants: both healthy youths and youths with ADHD (Finger et al., 2008
; Marsh et al., 2008
). These studies found amygdala or orbitofrontal cortex pathophysiology only in youths with psychopathic traits, not in youths with ADHD. Rubia and colleagues have also demonstrated that youths with CD who do not present with ADHD exhibit orbitofrontal cortex dysfunction whereas youths with ADHD do not (Rubia et al., 2009b
). Moreover, they have observed dysfunction in the recruitment of regions of lateral frontal cortex in youths with ADHD that was not seen, as it was not seen in the current study, in youths with pure CD (Rubia et al., 2008
; Rubia et al., 2009a
; Rubia et al., 2009b
). We therefore considered it unlikely that the group differences hypothesized in the current study would reflect ADHD and did not include this second comparison group.
In summary, the results of the present study indicate that psychopathic traits, which include remorselessness, shallow emotions, lack of empathy, and manipulativeness, and which predispose adolescents to severe ongoing antisocial behavior, are associated with atypical activity in the amygdala and orbitofrontal cortex during a moral decision-making task. We suggest that psychopathic traits may affect adolescents’ ability to attach the appropriate affective valence to actions of varying moral permissibility, and from using information about valence to guide their decisions.