Mood-Driven Risk Preference: How Induced Mood Affects Risk-Sensitive Learning

Abstract

How does mood influence one’s preference for risk through experiential learning? Mood has been shown to color various aspects of cognition, including the processes of learning and decision-making. Previous work has revealed that these very processes are also highly sensitive to risk – the variance associated with an outcome. While many studies have investigated the relationship between mood and risk-taking tendencies, this relationship in the context of trial-and-error learning has been underexplored. Drawing from theoretical and experimental findings, we propose that mood affects risk sensitive learning through nonlinear effects on the learning of probabilistic stimuli. To test this hypothesis, we recruited and tested 150 subjects on Amazon Mechanical Turk using a risk-sensitive reinforcement learning task containing experimental mood inductions (happy, sad, or neutral). We addressed the following research aims: (1) to examine mood’s effects on the learning of deterministic vs. probabilistic stimuli, (2) to compare distinct computational cognitive models of risk-sensitive learning, and (3) to tease out the mechanism by which mood drives risk preferences within the framework of the best-fitting model. Our behavioral results demonstrate a significant link between mood and risk attitudes, with a happy induction, relative to a sad induction, predicting a greater preference for risk. While the specific mechanism by which mood modulates risk preference is unclear, our results suggest the possibility of a more nuanced, dynamic model with mood in interaction with asymmetric learning.