UCLA

Despite mounting evidence that prosocial behavior is related to positive health outcomes, current understanding of prosocial neurodevelopment is limited. However, particular features of adolescent development (e.g., increases in perspective-taking, focus on peers) may facilitate prosocial behavior during adolescence, especially toward friends. The present work examines prosocial neurodevelopment from childhood through young adulthood by utilizing a reinforcement learning framework to assess prosocial behavior toward a friend. Participants (9-20 years old) completed self-report measures of empathy and selected a close friend. They then completed a multi-part reinforcement learning paradigm, including a) a learning phase while undergoing a functional magnetic resonance imaging scan, where performance was related to monetary outcomes for either the participant (self learning) or the selected friend (prosocial learning), b) a test phase to assess learning, and c) a surprise memory test to examine episodic memory to reinforcement events. A subset of participants completed a blood draw to assess the IFN-g, a cytokine related to antiviral immunity, a week later. Prosocial learning accuracy tended to peak in mid-adolescence, while the trajectory across age for self learning plateaued starting in early adolescence. Although children performed worse in terms of learning accuracy relative to other age groups, they had better episodic memory for the reinforcement events during learning. Hippocampal activity during learning was negatively associated with age and positively correlated with performance on the memory task, in line with the past work showing that the hippocampus plays a major role in memory. In females, empathy was differentially associated with prosocial learning performance across age, such that greater empathy was linked to worse prosocial learning accuracy in young adult females. This counterintuitive finding may reflect increased behavioral reactivity to other-relevant feedback in highly empathic individuals, leading to more volatile behavior and thus worse overall performance. No differences in neural activity were found with respect to prosocial vs. self learning. The subtlety of the manipulation may have contributed to the observed null results for comparisons between prosocial vs self learning in the present study. IFN-g was not significantly related to any study outcomes.