The impact of glucose administration on Bayesian v. heuristic based choice

We examine the impact of glucose on a Bayesian choice task that creates a separating equilibrium between high-level Bayesian choice and lower-level reinforcement heuristic choice. Consistent with a dual systems framework, we hypothesize that glucose administration will both increase reaction times and improve Bayesian accuracy because it should shift decision making towards the more deliberate system 2 and away from the more automatic system 1 decision process. We study 113 subjects randomly assigned to either a glucose or placebo drink condition, who make choices over several incentivized easy and difficult choices of the Bayesian task. Our results indicate a significant glucose effect on reaction times. In this case, glucose administration has a main effect of increasing reaction time, as predicted, but glucose also improves the marginal decrease in reaction times experienced across trials. Regarding Bayesian accuracy, we find that glucose administration significantly increases the likelihood of Bayesian choice over reinforcement heuristic-based choice only for those subjects indicating an above-average comprehension of the task structure. Additionally, we find that Bayesian choice likelihood increases across trials on the easy task when administered glucose. We interpret this, as well as the reaction time result, as evidence that glucose may improve learning, particularly on easy tasks. Together, these results suggest that there is a beneficial impact of glucose on deliberative decision making, though some of the results depend on the difficulty of the task and the comprehension of the decision environment being faced. Key Words: Bayesian Choice, Glucose, Learning, Experiments