Self-organization as a Key Principle of Adaptive Intelligence

Self-organization is a major functional principle in the living brain. This keynote develops on theoretical underpinnings of this concept to illustrate why it is also the mechanistic foundation of all forms of adaptive learning and intelligence, from the synapse to integrated neural networks, which are present in a large range of species from mollusks to the human primate. Starting with the single-synapse model of Hebbian learning, it is then clarified why the neurobiology, psychology and mathematics of self-organization deliver powerful models of knowledge generation across hierarchically organized, from-simple-to-complex, levels of functional integration. As a basis of adaptive intelligence, from sensory to cognitive learning and representation, the specificity, modular connectivity, and plasticity of biological neurons and networks allow function to grow independently. Without increasing structural system complexity, self-reinforced activity-dependent learning achieves stable representation while minimizing computations. Insights into the development of brain self-organization across evolution (phylogenesis) and a single human individual’s lifespan (ontogenesis) permit establishing the often missing functional link between mental and physical processes by clarifying how physical properties are encoded, acted upon, and transformed by adaptive intelligence. How novel solutions for robotics, where the activity of connections directly determines performance without necessity to add control functions, can arise from principles of self-organization is brought forward.