Bayesian Dissuasion with Bandit Exploration
arxiv(2024)
摘要
We investigate a two-period Bayesian persuasion game, where the receiver
faces a decision, akin to a one-armed bandit problem: to undertake an action,
gaining noisy information and a corresponding positive or negative payoff, or
to refrain. The sender's objective is to dissuade the receiver from taking
action by furnishing information about the payoff. Our findings describe the
optimal strategy for the amount and timing of information disclosure. In
scenarios where the sender possesses knowledge of the receiver's first-period
action or observes a noisy public signal correlated with it, the optimal
strategy entails revealing information in the second period. If this alone
proves to be insufficient to dissuade the receiver from acting, supplementary
information is provided in the first period.
In scenarios where information must be provided without conditioning on the
receiver's first-period action, the optimal strategy entails revealing
information exclusively in the first period.
更多查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要