Strategies for exploration in the domain of losses

* equal contribution The explore-exploit dilemma arises whenever we must choose between exploiting a known option or exploring the unknown in hopes of finding something better [1]. Most previous work has focused on explore-exploit scenarios in the domain of gains where the goal is to maximize reward. In many real-world decisions, however, the primary objective is to minimize losses. While optimal decision strategies should be the same for both gains and losses, it is well known that human decision making can shift significantly when confronted with losses. In this study, we compared explore-exploit behavior of human participants under conditions of gain and loss, using a model that measures two distinct exploration strategies: a directed strategy, in which exploration is driven by information seeking, and a random strategy, in which exploration is driven by decision noise. We found that people use both types of exploration regardless of whether they are exploring for gains or losses and that there is quantitative agreement between the exploration parameters across domains. Our model also revealed an overall bias toward the lesser known option in the domain of losses, independent of directed exploration, indicative of uncertainty seeking. Based on these results, we built a second, reduced model that explicitly distinguished information seeking from uncertainty seeking that was better able to explain the data. Taken together, our results show that explore-exploit decisions in the domain of losses are driven by three independen