We introduce a novel framework to enhance our understanding of behavioral heterogeneity. At the core of our investigation is the question of when and how individual behavior can be accurately inferred from aggregate data, particularly when individuals are susceptible to mistakes. Addressing this challenge requires bridging the gap between unobserved individual choices and observed aggregate outcomes. To this end, we first present the “bounds” framework, designed to constrain deviations from rational choice systematically. Then, we establish com- prehensive results on “decomposability,” a key property ensuring that aggregate choice behavior accurately mirrors the underlying characteristics of individual decision-making within a population.