Evidence demonstrating working memory limitations in human decision-making


If you were asked to evaluate the lottery [0.8765 probability of $99.98 or 0.1235 probability of $53.46], you will likely “edit” the numbers by rounding them (Kahneman & Tversky, 1979, Prospect Theory). Sims (1998, 2003, 2011) proposed the rational inattention hypothesis, positing that human computations in decision making are limited by finite Shannon capacity in working memory. We examined human performance in three conjunctive probability gambling tasks and found that subjects typically used 4-bit representations of probabilities in multiplying—evidence in support of Sims’ hypothesis.