CCN meeting | Andrea Reiter (Julius-Maximilians-Universität Würzburg, Germany), invited by Senne Braem

A reinforcement learning perspective on adolescent socio-emotional development

Adolescence is a key developmental window in which socio emotional behavior becomes increasingly shaped by dynamic interactions between affective states, social feedback, and learning processes. In this talk, I present a reinforcement learning perspective on both typical and atypical socio emotional development. I will showcase findings from three empirical studies. In the first part, I will talk about a study where we examined how social feedback modulates self worth learning in healthy adolescents and in adolescents with Borderline Personality Disorder. In this set of studies we showed that sensitivity to social prediction errors peaks in late adolescence and that young people with BPD fail to learn that they are liked. In the second part, I will talk about a study where we investigated mood–learning interactions in adolescents performing a reward based learning task. Our results show that moment to moment mood fluctuations systematically bias value updating and choice behavior, and that influence of such a mood bias peak in late adolescence, suggesting that affective states serve as a dynamic context for reinforcement learning during this developmental period. Together, this line of research illustrates the interplay between emotion, social experience, and learning in adolescence, refining our understanding of both vulnerability and resilience in socio emotional development.