Managing and arbitrating behaviours, processes and components in multilayered cognitive architectures when a huge amount of environmental variables are changing continuously with increasing complexity, ensue in a very comprehensive task. The presented framework proposes an hybrid cognitive architecture that relies on subsumption theory and includes some important extensions. These extensions can be condensed in inclusion of learning capabilities through bio-inspired reinforcement machine learning systems, an evolutionary mechanism based on gene expression programming to self-configure the behaviour arbitration between layers, a co-evolutionary mechanism to evolve behaviour repertories in a parallel fashion and finally, an aggregation mechanism to combine the learning algorithms outputs to improve the learning quality and increase the robustness and fault tolerance ability of the cognitive agent. The proposed architecture was proved in an animate environment using a multi-agent platform where several learning capabilities and emergent properties for self-configuring internal agent's architecture arise.