Decision, Vol 11(3), Jul 2024, 383-403; doi:10.1037/dec0000225
Past research on reward-based decision making documents deviations from optimal behavior that suggest insufficient sensitivity to rare outcomes. This insensitivity to rare events has often led to suboptimal performance in gambling tasks. Participants tend to select options that provide frequent rewards, even though these same options sometimes give large-magnitude losses. In this study, we tested an experimental manipulation designed to enhance attention to the magnitude of choice outcomes, where outcomes were presented in segments of 100 points rather than all at once. Performance in the segmented condition was compared to performance in the original condition, where reward amounts were not segmented. Participants in the segmented condition performed better than participants in the original condition in both a standard gambling task, involving both gains and losses, and a gains-only task where no losses were given. A third experiment indicated that the increase in instrumental responses, or button presses, for reward outcomes, was a critical factor that led to better performance in the reward segmentation conditions. We propose that the reward segmentation manipulation caused trials with rare, large-magnitude rewards to be perceived as additional trials associated with the rare option, which eliminated participants’ underweighting of rare events, leading to better performance. These results suggest that segmenting the presentation of rewards is a potential way in which decision-making behavior can be improved, particularly in cases where some outcomes are rare. (PsycInfo Database Record (c) 2024 APA, all rights reserved)