Abstract
Outcomes of past decisions profoundly shape our behavior. However, choice-outcome associations can become volatile and adaption to such changes is of importance. The present study combines pharmaco-electroencephalography with computational modeling to examine whether intranasal oxytocin can modulate reinforcement learning under a volatile vs. a stable association. Results show that oxytocin increases choice accuracy independent of learning context, which is paralleled by a larger N2pc and a smaller P300. Model-based analyses reveal that while oxytocin promotes learning by accelerating value update of outcomes in the volatile context, in the stable context it does so by improving choice consistency. These findings suggest that oxytocin's facilitatory effects on learning may be exerted via improving early attentional selection and late neural processing efficiency, although at the computational level oxytocin's actions are highly adaptive between learning contexts. Our findings provide proof of concept for oxytocin's therapeutic potential in mental disorders with adaptive learning dysfunction.