Proceedings of the Annual Meeting of the Cognitive Science Society

Risk communication often relies on descriptive probabilities, yet there is a general preference for experience-based impressions that can be biased by small samples. Interactive simulations can combine both modes, but evidence is mixed and key design features remain confounded. In two online experiments (N = 422; N = 360), we tested how features of simulated experience affect expected-value maximization (EV-Max) in monetary loss gambles. Probabilities were communicated using icon arrays only, and each option involved two independently co-occurring losses. In Experiment 1, the presentation format had only marginal effects, with no advantage for active sampling, whereas sampling without replacement robustly increased EV-Max choices. In Experiment 2, sampling control did not affect EV-Max choices; accumulative displays yielded only marginal EV-Max gains but increased EV-Max preference, confidence, perceived informativeness, and search length. Overall, increased autonomy was largely irrelevant, while reducing sampling noise and using accumulative summaries appear beneficial for risk communication.