|
Project Information
|
Can artificial agents cooperate like humans in a public good experiment?An attempt to get q-learners with neural networks to create their own institution given punishment and a "definition" of the rule. The rule is that everyone contributes their endowment and punishes those who do not contribute. Operationally, this definition is the ability to perceive that others are following the rule to less or more of a degree. Once we have results, we will compare them to human subject data. This research project uses pybrain. See http://pybrain.org. |