Learning, experimentation and the convergence of the discovered preferences

Abstract

In this article I study whether the interim preferences of the consumer can be expected to converge to their real preferences in the process of preference discovery. I construct a subjective expected utility model of the consumer, where the uncertainty results from the imperfect knowledge of their own preferences. This uncertainty is partially resolved by experimental consumption. Under the assumption that the subjective probability of the consumer satisfies learning monotonicity, I identify the equivalent conditions for the consumer to experiment. My results show that the interim preferences never fully converge to the real preferences of the consumer. Instead, the preference discovery either terminates, meaning that the consumer ceases to experiment, or only experiments within some neighborhood of the best currently known alternative, and never sufficiently explores their preferences.