How bad becomes good: A neurocomputational model of flexible affect valuation

Although people often prefer to pursue pleasant affect, in many situations, unpleasant affect is more valuable. Yet little is known about how people flexibly value different affective valences across contexts. Do the same neural circuits that generate affect generate value? What differentiates people who more flexibly value affect? To investigate these questions, we developed a neurocomputational model of affect valuation, in which people convert subjective affect into context-sensitive decision value through a process of weighted evidence accumulation. We then tested model predictions by recording EEG and facial EMG during a novel affective choice paradigm in a sample of racially diverse, undergraduate participants (data collected in 2018-2019). Consistent with the model, we found that generation of affective responses occurs earlier than, and is neurally distinct from, valuation of that affect. Moreover, individual differences in flexibly valuing affect correlated only with later valuation processes, not earlier affect generation processes. Our results have important theoretical implications for emotion, emotion regulation, and decision making.