Control over Sampling and Generation Process: Features of Experience-Based Learning

Effective risk communication depends on how probabilistic information is presented, especially in applied contexts with multiple outcomes. This study examined how features of simulated experience formats—presentation format (static, passive, or active sampling) and sampling generation process (with or without replacement)—influence risk perception and decision making. In an online experiment (N = 422), participants viewed visual probability representations for two options, each with two independent outcomes, and selected their preferred option. The main outcome was adherence to the expected value maximization strategy. We also measured preference, subjective values, probability estimates, confidence, and perceived informativeness. Passive sequential sampling without replacement produced the highest adherence to expected value maximization. Static formats yielded more accurate probability estimates and were rated as more informative. Active sampling offered no additional benefits and may have imposed extra cognitive load. About half of participants followed expected value principles, highlighting the importance of structured sampling in probabilistic reasoning.