Development of an Anticipating Decision Making System Using Evolution and Learning

The target of this thesis was to develop an anticipating and learning deci- human decision sion making system for competitive decision problems. Humans are so- making as model phisticated decision makers - they are not purely stimulus-driven, but internally simulate possible strategies and consequences before their execution. To allow the integration of environmental changes into this simulation (e.g. represented by a dynamic opponent in a competitive interaction), an imitator model is built. This allows to adapt the different available decision strategies, which humans have in mind, to a dynamically changing environment. After an evaluation according to personal preferences the resulting decision is made. These decision making concepts allow humans to make very exible and robust decisions. Neurobiological ndings cannot provide detailed guidelines for the development of a system. Nevertheless hypotheses regarding the general architectural approach can be translated into arti cial systems.