Children and adolescents with conduct disorder (CD) (1
) show a propensity toward risk taking and reckless behavior, suggesting difficulties with decision making and impulsivity. They are also more susceptible to substance abuse (2
), potentially reflecting altered sensitivity of reward mechanisms and persistent selection of options with short-term benefits despite negative long-term consequences.
Decision-making difficulties in CD may stem from deficits in emotional and executive function (EF). Bechara et al.
) have demonstrated that patients with ventromedial frontal cortex damage exhibit decision-making impairments on the Iowa Gambling Task (IGT) (3
), interpreting these results as reflecting an inability to use somatic or emotional markers to choose between competing options and, in particular, signify options with potential for adverse consequences (4,5
). Conduct disorder is associated with reduced amygdala and anterior insula volume (6
) and neuropsychological deficits consistent with putative amygdala dysfunction (7–9
). These deficits may alter the influence of emotion on decision making in individuals with CD. Evidence for executive dysfunction in CD or its milder variant, oppositional defiant disorder (ODD), is less compelling, although children with ODD experience difficulty in responding to changes in environmental contingencies, e.g., suppressing responses to previously rewarded stimuli when they become associated with punishment (10
Given these lines of evidence, decision-making impairments might be expected in individuals with CD, but empirical data are limited. Compared with control subjects, adolescents with disruptive behavior disorders (DBDs) only exhibited deficits in IGT performance when playing it for a second time, a week after the first attempt (11
), thereby failing to show improved performance over time.1
Iowa Gambling Task performance was significantly impaired in children with high levels of psychopathic traits, relative to control subjects (12
). It is not known whether CD participants low in psychopathic traits show decision-making impairments.
The first aim of the study was to characterize decision making under risk, where outcome probabilities are explicitly provided, under differing motivational conditions in male adolescents with CD and healthy control subjects. Iowa Gambling Task performance deficits may reflect impairments in multiple neuropsychological processes, including working memory, reversal learning, or sensitivity to reward/punishment (13
). Consequently, we used a modified version of the Risky Choice Task (RCT) (14
), a more direct measure of decision making that could be played twice in the test battery to examine effects of increased motivation and stress. We felt it would be informative to measure decision making under conditions of heightened motivation and stress because differences between control subjects and individuals with DBDs may be most evident in “hot” motivational contexts (10,15
). This could be partly due to physiological hyporeactivity observed during stress in children and adolescents with DBD, particularly CD and ODD (16,17
). Furthermore, by enhancing participants' motivation to perform well on the task, we sought to minimize the possibility that apathy or lack of engagement would underlie group differences in decision making.
A second aim was to assess global executive function in CD using the Wisconsin Card Sorting Test (WCST) (18,19
). This would allow us to assess the relationship between global executive function and decision making and test the specificity of changes in motivational or emotional aspects of executive function that would be reflected in RCT performance changes. Among other cognitive processes, the WCST measures set shifting, which can be considered a measure of “cold,” or nonaffective, executive function (20
). Neuroimaging studies suggest that the RCT and WCST may recruit partially dissociable neural circuits (21,22
), consistent with this fractionation of hot and cold executive function.
Our third aim was to examine potential effects of age of CD onset, as early-onset (or childhood-onset) CD is suggested to be uniquely associated with neuropsychological impairment (23
). In contrast, adolescence-onset CD is argued to arise primarily due to social modeling of antisocial peers. We investigated whether this distinction, as suggested by the developmental taxonomic theory (23
), would extend to differences in decision making and reward mechanisms. Previous studies reported intact WCST performance in male early-onset and adolescence-onset CD participants (24
), although female participants with CD showed increased perseverative errors on the WCST (25
). However, few data are available on decision-making processes or hot executive function in CD.
We hypothesized that increased motivation/stress would promote cautious choices on the RCT in control subjects. Adolescents with early-onset CD were predicted to show increased risky decision making relative to control subjects, with differences most pronounced under conditions of increased motivation/stress. We anticipated group differences in experimental gamble frequency following certain outcomes, with CD participants less dissuaded by losses in previous trials due to insensitivity to punishment cues. Given their impaired function, vulnerability to substance abuse, and poorer adult outcomes (26
), we also expected heightened risky decision making in adolescence-onset CD participants, contrary to the developmental taxonomic theory (23
). Finally, early-onset CD participants were predicted to show increased WCST perseverative errors given previous data showing response perseveration in children with ODD (10
) and female adolescents with CD (25