主题:How Actions Rewire Preferences: A Normative Bayesian Model Bridging Lab and Large-Scale Reality

时间:  2026年4月29日 下午15：00

地点: 管理学院一楼第四教室

主讲人: 殷云露，复旦大学

主持人: 孙浩

Bio:

殷云露，复旦大学管理学院市场营销学系副教授。博士毕业于香港大学经管学院。研究兴趣主要集中在视觉营销、媒体与数字营销，人机交互，以及决策神经科学。研究成果发表在Journal of Consumer Research, Journal of Marketing Research, International Journal of Research in Marketing, Journal of Business Ethics, Journal of Retailing, Elife, Neuroimage 等营销学与认知神经科学期刊上。目前担任Journal of Business Research副主编，以及JMR、JCR等期刊审稿人。研究获教育部第九届高等学校科学研究优秀成果奖（人文社会科学）青年成果奖, 并入选上海市浦江人才计划、上海市领军人才（海外）计划，并主持自然科学基金项目两项。

照片：

Abstract: 

Whether and how preferences evolve after choices has been a long-standing open question in the field of decision-making. While classic economic theories assume stable preferences, decades of empirical studies have shown that preferences will change systematically after selecting or rejecting an option. Yet it remains unclear when, how, and why such changes occur. Here, we propose a Bayesian Preference Change (BPC) model that formalizes an individual’s internal preference as a probabilistic belief, which is updated through Bayesian inference after each decision. The model not only explains the effects of action-induced preference change but also generates a set of testable predictions for how preference will change. We tested the BPC model in a laboratory behavioral experiment (N = 95) and applied the model-derived insights to real-world shopping data, consisting of 30,153,078 purchase transactions by 2,395,651 consumers from a major retail chain. For the lab experiment, the BPC model closely captured post-choice rating changes and outperformed alternative models for most participants. All model-derived predictions were robustly supported. For large-scale behavioral traces, simple behavioral markers derived from the BPC model can identify individuals with higher future purchase growth. Together, these findings demonstrate that action-induced preference change reflects Bayesian updating, offering a normative explanation for this longstanding yet seemingly irrational phenomenon. More importantly, the BPC model goes beyond the lab, yielding theoretical insights that explain large-scale human behavior.