How sampling strategies shape experience-based risky choice

Hallmark phenomena of risky choice, such as risk aversion and deviations from expected value (EV) maximization, are commonly modeled with psychoeconomic curves (e.g., utility function, probability weighting function). Yet these functions describe choices rather than the cognitive processes that generate them. Here, we examine how aspects of the oft-neglected process of information search - in particular, switching between options during sampling and stopping based on sampled evidence - can give rise to patterns in experience-based risky choice in the context of the sampling paradigm. We develop a computational framework that conceptualizes sampling strategies as consisting of three building blocks: a search rule (governing the rate of switching between options during sampling), a comparison rule (roundwise vs. summary evaluation of the options), and a stopping rule (decision threshold). In simulation analyses comparing different combinations of these building blocks, we show that frequent switching increases EV-maximizing choices for strategies with a summary comparison rule, but decreases maximization for strategies with a roundwise comparison rule. Further, frequent switching promotes an apparent underweighting of rarely experienced events found under the roundwise comparison rule, and curtails overweighting found under the summary comparison rule. We also reveal how the sampling strategies give rise to different risk attitudes depending on the existence and attractiveness of rare events. Finally, an empirical analysis suggests that people combine switching behavior and comparison rules in a way that fosters EV-maximizing choices. Our results underscore the possible contribution of search processes to patterns in risky choice.