Fast or Rational? A Response-Times Study of Bayesian Updating
基于心理学双过程理论，构建了一个决策模型，其中直觉过程依赖强化学习，理性过程依赖贝叶斯更新。实验发现，当两个过程冲突时，正确反应更慢；一致时，错误反应更慢。
We present a simple model for decision making under uncertainty building on dual-process theories from psychology, and use it to illustrate a possible component of intuitive decision making of particular relevance for managerial settings. Decisions are the result of the interaction between two decision processes. The first one captures optimization based on Bayesian updating of beliefs. The second corresponds to a form of reinforcement learning capturing the tendency to rely on past performance. The model predicts that (i) in the case of conflict between the two processes, correct responses are associated with longer response times, but (ii) if both processes are aligned, errors are slower. Furthermore, (iii) response times in the case of conflict are longer than in the case of alignment. We confirm the predictions of the model in an experiment using a paradigm where an associative win-stay, lose-shift process conflicted with rational belief updating. Data, as supplemental material, are available at http://dx.doi.org/10.1287/mnsc.2013.1793 This paper was accepted by Yuval Rottenstreich, judgment and decision making.