Learning by consuming in the lab

Abstract We build an experiment to uncover the bandit-like nature of consumer behavior usually masked by the product and time aggregation of consumption data. Subjects make repeated choices between musical styles (either all familiar or unfamiliar), and post-choice satisfaction is observed. We estimate Bayesian bandit models of learning taste by consuming with satiation. Our best model features decreasing random exploration, with openness being associated with higher exploration. Early exploration is more intensive in the unfamiliar treatment and persists throughout the experiment in both treatments. Overall, subjects make choices that deviate from their best prediction 61.5% of the time in the unfamiliar treatment versus 44.7% in the familiar treatment. Our model offers a rational interpretation of random utility in discrete consumer choices which does not rest on perception and/or decision errors.