Thesis: Beliefs and preferences are two building blocks which, together with the concept of rationality, underlie most descriptive and normative analysis in economics. This dissertation explores the (old) idea that beliefs and preferences are not fixed, but instead vary in reaction to the environment. Chapter 1, joint with Agathe Pernoud, studies informational incentives in mechanism design, that is how the institutional environment affects how agents acquire information. Agents have uncertainty on their preferences and can buy information on their valuations as well as others’. We show that, under certain conditions, agents acquire information on others’ whenever the mechanism violates a separability condition which rules out most economically meaningful mechanisms. Chapter 2, joint with Philippe Jehiel, studies how students form their subjective admission chances. We propose a model in which students average the past experience of their peers which induces a regression to the mean of admission chances. Two inefficiencies arise: high-achieving disadvantaged students self-select out of elite colleges, and low-achieving advanteged students wastefully apply to elite colleges even though their admission chances are zero. Chapter 3, joint with Niels Boissonnet and Alexis Ghersengorin, investigates the question of preference change. We show that it is possible to falsify and identify a model of preference change in which the decision maker changes her behavior according to a meta-preference relation. Chapter 4, joint with Agathe Pernoud, studies how beliefs about economic models affect strategic communication, for instance between voters and medias. We show that holding a misspecified model can increase the informativeness of communication, and model misspecification is stable as communication on models by a third party is generally limited.