Design and Integration of Partial Brain Models Using Hierarchical Cooperative CoEvolution

The current work addresses the problem of designing and integrating brain-inspired artificial cognitive systems. Specifically, we introduce a new computational framework for modelling partial brain areas following a coevolutionary agent-based approach. Properly formulated neural agents are employed to represent brain areas. A cooperative coevolutionary method, with the inherent ability to co-adapt substructures, supports the design of the models, and additionally provides a consistent methodology to achieve their integration. The implemented models are successfully embedded in a simulated robotic platform which supports environmental interaction. The proposed approach is utilized to design two distinct models: one for primary motor cortex able to drive the robot in a purposeless wall avoidance mode, and one for hippocampus which supports selflocalization. These models are further integrated by adding at the same time prefrontal structures, in order to drive the robot in a purposeful mode, accomplishing a DMS task in a cross (+) maze.