Abstract
A theoretical framework is developed to analyze individual bracketing behavior in risk aggregation tasks. The methodology is applied in a behavioral experiment with real financial incentives, using pairs of independent binomial gambles. Results reveal substantial variation in bracketing behavior across subjects, unrelated to demographics but linked to gamble risk profiles. A minority systematically applies Broad or Narrow Bracketing. In contrast, most subjects display Partial Bracketing with high variability, suggesting choice inconsistency rather than a systematic rule aimed at reducing cognitive burden or maximizing hedonic value.