Decision-making under uncertainty is pervasive. This work sought to understand the role of working memory (WM) in loss sensitivity by utilizing two widely used tasks, the Iowa Gambling Task (IGT) and the Soochow Gambling Task (SGT), and manipulating WM with a dual-task paradigm. We hypothesized that WM load would reduce attention to both loss value and frequency in the decision-making tasks. To better delineate the psychological processes underpinning choice behavior, we developed an Expectancy-Frequency-Perseveration (EFP) model which parsimoniously captures three critical factors driving choices: expected value, frequency of gains and losses, and perseveration. Behavioral and computational modeling results indicate that WM load compromised performance in the IGT due to reduced attention to loss value but enhanced performance in the SGT because of diminished attention to loss frequency. Our findings suggest that WM heightens attention to losses, but that greater attention is given to loss frequency than loss value.