The aberrant RSFC of the amygdala network and the dACC network may be related to altered emotion processing and regulation in depressed adolescents. Our results provide new insights into RSFC in clinically depressed adolescents and future models on adolescent depression may include abnormalities in the connectivity of salience network.

Prior developmental functional magnetic resonance imaging (fMRI) studies have demonstrated elevated activation patterns in the amygdala and prefrontal cortex (PFC) in response to viewing emotional faces. As adolescence is a time of substantial variability in mood and emotional responsiveness, the stability of activation patterns could be fluctuating over time. In the current study, 27 healthy adolescents (age: 12-19 years) were scanned three times over a period of six months (mean test-retest interval of three months; final samples N=27, N=22, N=18). At each session, participants performed the same emotional faces task. At first measurement the presentation of emotional faces resulted in heightened activation in bilateral amygdala, bilateral lateral PFC and visual areas including the fusiform face area. Average activation did not differ across test-sessions over time, indicating that at the group level activation patterns in this network do not vary significantly over time. However, using the Intraclass Correlation Coefficient (ICC), fMRI reliability demonstrated only fair reliability for PFC (ICC=0.41-0.59) and poor reliability for the amygdala (ICC<0.4). These findings suggest substantial variability of brain activity over time and may have implications for studies investigating the influence of treatment effects on changes in neural levels in adolescents with psychiatric disorders.

Learning from feedback lies at the foundation of adaptive behavior. Two prior neuroimaging studies have suggested that there are qualitative differences in how children and adults use feedback by demonstrating that dorsolateral prefrontal cortex (DLPFC) and parietal cortex were more active after negative feedback for adults, but after positive feedback for children. In the current study we used functional magnetic resonance imaging (fMRI) to test whether this difference is related to valence or informative value of the feedback by examining neural responses to negative and positive feedback while applying probabilistic rules. In total, 67 healthy volunteers between ages 8 and 22 participated in the study (8–11 years, n = 18; 13–16 years, n = 27; 18–22 years, n = 22). Behavioral comparisons showed that all participants were able to learn probabilistic rules equally well. DLPFC and dorsal anterior cingulate cortex were more active in younger children following positive feedback and in adults following negative feedback, but only when exploring alternative rules, not when applying the most advantageous rules. These findings suggest that developmental differences in neural responses to feedback are not related to valence per se, but that there is an age-related change in processing learning signals with different informative value.