040065 UK Modelling bounded rationality (MA) (2020W)

Rationality in economics typically refers to predictions based on expected utility maximization. Subjects need to know all their options, need to know all possible consequences of any of their choices and have to assign probabilities to any event that they are uncertain about. In this course we are interested in methodologies that do not make these heroic assumptions but satisfy some principles that identify behavior as being optimal in some sense. Behavior is optimal in a minimal sense if it can be backed up by a reason for acting. We will not limit ourselves to this minimal interpretation of optimality as this will not give us any insight as to whether we should expect such behavior to survive over time. The presumption is that a heuristic as a model of how to make choices can only emerge if it has properties that signal it out as being optimal or particularly well adapted to a situation. A central paper in this course (Schlag, 1998) was recently elected as one of the 50 most influential papers in a leading economic theory journal in the past 50 years. The objective therein is to understand how individual can learn from each other with only very limited understanding of the consequences of their behavior. It is the way that individuals learn which can make them very smart as a whole. We will also consider individuals who can only learn from their own past experiences. Buzz words include imitation, cultural evolution, heuristics, tabus and reinforcement learning. A central focus of the course will be on fostering a discussion about the philosophical and economic approaches to rationality and how the latter can be relaxed.