When and why rare events are underweighted: a direct comparison of the sampling, partial feedback, full feedback and description choice paradigms.

Two paradigms are commonly used to examine risky choice based on experiential sampling. The feedback paradigm involves a large number of repeated, consequential choices with feedback about the chosen (partial feedback) or chosen and foregone (full feedback) payoffs. The sampling paradigm invites cost-free samples before a single consequential choice. Despite procedural differences, choices in both experience-based paradigms suggest underweighting of rare events relative to their objective probability. This contrasts with overweighting when choice options are described, thereby leading to a 'gap' between experience and description-based choice. Behavioural data and model-based analysis from an experiment comparing choices from description, sampling, and partial- and full-feedback paradigms replicated the 'gap', but also indicated significant differences between feedback and sampling paradigms. Our results suggest that mere sequential experience of outcomes is insufficient to produce reliable underweighting. We discuss when and why underweighting occurs, and implicate repeated, consequential choice as the critical factor.