One approach to modeling economic choices made under bounded human rationality is to develop theoretical economic agents that act and choose as actual humans do. Agents could be represented as using parametrized decision algorithms, and these algorithms could be chosen and calibrated so that the agents' behavior matches real human behavior observed in the same decision context. A learning algorithm is developed and calibrated for of a simple decision context, that of agents choosing repeatedly among discrete actions with initially unknown, random consequences. The analysis indicates that artificial learning agents can be designed and their rationality calibrated to replicate human behavior. The agents' behavior reproduces 2 stylized facts of human learning: 1. With frequency-dependent payoffs, humans meliorate rather than optimize. 2. There is a threshold in discrimination among payoffs below which humans may lock in to suboptimal choices.
Mendeley saves you time finding and organizing research
Choose a citation style from the tabs below