Do people care about loss probabilities?

In a series of experiments, we provide evidence that people pay special attention to the probability of losing. We first analyze this behavior in the typically used one-shot choice tasks. We then extend our analysis to repeated decisions in choice tasks, as well as allocation and investment tasks. Additionally, we test both decision making under risk and under gradually removed uncertainty, as with decisions from experience. Our findings of explicit attention to loss probabilities contradict the predictions of normative and descriptive decision theories, such as Expected Utility Theory and (Cumulative) Prospect Theory. We suggest a value function with a jump rather than a kink at the reference point, which separates gains and losses.