This symposium brings together four contributions exploring the development of behavioral adaptation under uncertainty from childhood to adulthood, utilizing fMRI, computational modeling, and ecological momentary assessment.

Maria Waltmann presents findings from an fMRI study on probabilistic reversal learning in adolescents with and without childhood maltreatment. Maltreated adolescents exhibit quicker adaptation in win-seeking contexts, with enhanced perception of environmental volatility and significant involvement of the right dorsolateral prefrontal cortex, suggesting neurocognitive adaptations to unpredictable environments.

Johannes Falck discusses the impact of individual (e.g., body-mass-index BMI) and environmental (familial psychopathology, socio-economic status SES) risk factors on reinforcement learning in a developmental study. Higher BMI and lower SES predict parameter differences, while familial psychopathology and low SES moderate symptom-parameter relationships, influencing learning outcomes and decision-making.

Gabriele Chierchia examines the development of strategic social decision-making in adolescents through a tacit coordination task. Adolescents increasingly adapt decisions to cooperative and competitive incentives, with significant developmental changes in decision variability and response times, highlighting adolescence as a critical period for developing strategic social abilities.

Klara Gregorova explores the influence of social feedback on self-esteem across age groups and clinical populations. Using an experimental paradigm and ecological momentary assessment, the study shows self-esteem reactivity to social feedback increases during adolescence and decreases in adulthood. Participants with Borderline Personality Disorder exhibit negatively biased expectations of being liked, despite receiving positive feedback.

Together, these contributions provide a comprehensive understanding of behavioral adaptation in children and adolescents, emphasizing their relevance for understanding mental health risk factors during development.

Childhood maltreatment significantly impacts mental health, but the underlying neurocognitive mechanisms remain unclear. While previous research shows that maltreatment alters threat and reward processing, its effects on value-based learning and decision-making are not well understood. We investigated how maltreatment influences learning and decision-making in volatile environments focused on win-seeking and loss-avoidance, respectively. Adolescents with (n=59) and without (n=46) maltreatment exposure completed a probabilistic reversal learning task with a win and a loss block during fMRI. Computational modeling using a Hidden Markov Model showed that, relative to non-maltreated peers, maltreated adolescents had lower estimates of environmental stability in the win condition, which led to enhanced behavioral flexibility and improved performance after reversals. This phenotype was accompanied by reduced right dorsolateral prefrontal cortex activation associated with upcoming reversals. Within the maltreatment group, lower estimates of stability in the win condition were correlated with psychiatric symptoms. Interestingly, despite the advantage they conferred after reversals, lower estimates were also linked to impaired performance during stable task phases in this group. Taken together, our findings thus challenge the view that maltreatment uniformly impairs cognitive function. Instead, they suggest that maltreatment may lead to specific adaptations that enhance cognitive-behavioral flexibility in win-seeking contexts, which can be advantageous in volatile environments. However, if excessive, these adaptations may lead to maladaptive decision-making in stable contexts, potentially contributing to psychopathology. Our study thus provides novel insights into how early adversity shapes reinforcement learning and decision-making at a neurocomputational level, with implications for understanding the development of psychopathology.

Mental health influences adaptive behavior during Reinforcement learning (RL), but the impact of individual and environmental risk factors during development remains unclear. We examined how body mass index (BMI), familial psychopathology (FHP), and socioeconomic status (SES) influence RL and moderate symptom-RL relationships. Using a cross-sectional developmental dataset (study 1, age 8-18, n=122, clinical and non-clinical individuals), we tested these effects and validated symptom-independent effects in two developmental non-clinical datasets (study 2, n=60, age 6-9, 13-16; study 3, n=141, age 6-7, longitudinal 2-year follow-up).

Higher BMI predicted lower reward learning rates in studies 1 and 2. In study 3, BMI predicted lower cross-sectional learning rates in girls under delayed feedback and lower learning rate gains longitudinally.

Lower parental SES predicted lower learning rates in study 1 (education, occupation status) but showed no effects in study 2. In study 3, lower SES predicted both lower cross-sectional learning rates (income), and higher learning rates in girls (occupation status), along with reduced longitudinal learning rate gains from delayed feedback in girls (income, occupation status).

Familial psychopathology did not predict RL parameters directly. Instead, the presence of psychopathology among first and second-degree relatives as well as lower SES predicted larger detrimental effects on RL parameters in relation to symptoms of depression, anxiety and ADHD.

In conclusion, while risk factors such as familial psychopathology and lower SES moderated symptom-RL parameter links, higher BMI and lower SES predicted RL parameter differences and longitudinal changes that largely align with effects of psychopathology symptoms.

Self-esteem is an important cornerstone of mental health across the lifespan. Throughout life, but perhaps particularly during adolescence, self-esteem is shaped by appraisals from other people. Previous work showed that momentary fluctuations in self-esteem are best explained not only by whether we are liked by others, but whether we are liked more or less than we expected. However, it is not known whether adolescents and adults, or adolescents with and without pathological instability (i.e., symptoms of Borderline Personality Disorder) differ therein. Therefore, the present study investigated age- and symptom-related differences in the extent to which self-esteem is shaped by social experiences. Study 1 included 141 typically developing participants (10 to 40 years) while study 2 included 55 patients with BPD symptoms and 69 healthy controls (13-25 years). We conducted an experimental paradigm that simulated peer evaluation in a social media context and complemented them by real-life Ecological Momentary Assessment. Across development, self-esteem reactivity in response to unexpected social feedback increased during adolescence and decreased during adulthood. Real-life self-esteem reactivity decreased with age in response to the pleasantness of previous social interactions. Both measures were related. In the clinical sample, patients reacted more strongly to feedback received from unkind raters, which was associated to dysfunctional parental styles. Beyond self-esteem reactivity, patients showed largely negatively biased expectations about being liked, which persisted and could not be corrected by positive feedback. These findings provide insights into the strong role of social determinants of self-esteem and the specificity of developmental versus pathological changes.