No Bad Dogs: Ethological Lessons for Learning in Hamsterdam*

摘要: We present an architecture for autonomous creatures that allows learning to be combined with action selection, based on ideas from ethology. show how temporal-difference may used within the context of ethologically inspired animat build and modify portions behavior network, set fundamental parameters including strength associated individual Releasing Mechanisms, time course appetitive behaviors, rates observed reliability specific contingencies. The algorithm has been implemented as part Hamsterdam toolkit building animated creatures. When in Silas, a virtual dog, enables Silas trained using classical instrumental conditioning.