Learning in a Post-Truth World
研究了错误信息如何影响两种经典社会学习模型（贝叶斯和DeGroot），发现认知更复杂的个体反而更容易受错误信息影响，对政策制定有警示意义。
Misinformation has emerged as a major societal challenge in the wake of the 2016 U.S. elections, Brexit, and the COVID-19 pandemic. One of the most active areas of inquiry into misinformation examines how the cognitive sophistication of people impacts their ability to fall for misleading content. In this paper, we capture sophistication by studying how misinformation affects the two canonical models of the social learning literature: sophisticated (Bayesian) and naive (DeGroot) learning. We show that sophisticated agents can be more likely to fall for misinformation. Our model helps explain several experimental and empirical facts from cognitive science, psychology, and the social sciences. It also shows that the intuitions developed in a vast social learning literature should be approached with caution when making policy decisions in the presence of misinformation. We conclude by discussing the relationship between misinformation and increased partisanship and provide an example of how our model can inform the actions of policymakers trying to contain the spread of misinformation. This paper was accepted by Omar Besbes, revenue management and market analytics.