ABSTRACT
The MicroPsi agent architecture is a step towards modeling human cognition and emotion and thereby gaining a better understanding. The cognition of MicroPsi agents is embedded into an emotional and motivational sub-system that modulates perception, memory retrieval and action control. MicroPsi agents are part of a larger framework, which has been developed by our group and includes a simulation environment and tools for setting up experiments, allowing for interaction and 3D visualization. In addition, a focus is laid on the motive and interaction based acquisition of representations of the environment, similar to the concept of affordances. On the level of cognitive mechanisms, it suggests a mode of representation that unifies connectionist and symbolic principles. The resulting formalism may be described as a Bayesian network with additional logic to implement spreading activation, hierarchical representation and script execution. In MicroPsi, these networks are used to express active and sensor schemata, event protocols, plans and control structures.
