Comparing the Description-Experience Gap Across Numerical and Pictographic Risk Representations

Depending on how information is conveyed, people show different risk preferences. Odds and outcomes could be described (e.g., Option A [90%: £20, 10%: £8] and Option B [certain £18]), where people often overweight rare outcomes [10%: £8]. However, when odds and outcomes are learned through experience (sampling), people tend to underweight rare outcomes, a phenomenon known as the Description-Experience Gap. Previous studies demonstrating these effects have relied on numerical risk representations. However, risk communication could also be visual, using pictographs, which may lead to different preferences due to the integration of elements from both description- and experience-based paradigms and pattern recognition processes. Across two pre-registered experiments, we examined the D-E Gap across numerical and visual risk representations. Participants made risky choices under four within-subjects conditions: (1) Numeric Description—full numeric summary of each lottery; (2) Pictograph Description—lotteries represented with arranged icon arrays; (3) Numeric Sampling—numeric outcomes sampled sequentially; and (4) Pictograph Sampling—outcomes sampled sequentially from icon arrays. We observed a reduced D-E Gap with visual formats—primarily driven by the Pictograph Description. In Experiment 2, we further tested how variations in these conditions, related to attenuating memory reliance (via a visible history of past samples) and sampling effort (via automated sampling), accounted for the between-condition differences. Our results persisted despite these changes—providing new insights into how distinct cognitive processes relate to this risk preference behaviour. Exploratory analyses indicate that observing of all outcomes during sampling had a substantial effect on the magnitude of the D-E gap.