Conference on Cognitive Computational Neuroscience

Economic choices are known to be strongly influenced by both outcome valence (i.e., whether an outcome entails gain or loss) and risk (the variance of outcomes). It has been shown in previous studies that people have different risk preferences when the prospects’ information is described explicitly and when they are learning it by experience, known as the experience-description gap. However, the effect of valence on behavioral performance in an experienced-based learning task has been less investigated. To address this question, we used a novel instrumental learning task in 2 online behavioral experiments designed to manipulate risk and valence independently. Consistent results in both experiments show a significant effect of a prospect’s riskiness on behavioral performance. Moreover, our results indicate during experience-based risky choices, people are more risk-seeking in gains and risk-averse in losses, which is in contrast with their preferences in a description-based risky choice and contrary to the reflection effect discussed in the prospect theory. Our results reveal the importance of context-dependent learning in shaping risk preferences in experience-based choices.