"What does the brain do?", or Solving a priori undefined problems based on an autonomous agent behavior in real environment

The relevance of the work is determined by the need to develop theoretical foundations for the creation of universal artificial intelligence systems.

Aim. This work is to create a conceptual model of the systemic essence of the brain.

Research methods. Of fundamental importance is that both ontology meta-matching technique and problem-solving algorithms synthesize points and segments along an agent'smovement in a behavioral space. This means that ontologization, identification, and solution of all problems encountered by the agent in the real environment during its existence are accomplished through the synthesis and implementation of the agent'sbehavior in that environment.

Results. A hypothesis about the systematic target function of the brain has been formulated, and a computational model of the basic processes underlying its operation has been developed, based on the analogy of designing a neurocognitive system to control the behavior of an autonomous agent in a real-world environment. A concept of an agent'sexistence as a trajectory in its behavioral space was proposed, which is a mathematical abstraction of the state space of the "agent – environment – observer" system. A concept for adapting agents to real-world conditions has been proposed, based on brain-implemented meta-algorithms for ontologization and algorithms for identifying and solving a priori undefined problems, grounded on a metaphor for designing an autonomous synthesis of the agent'strajectory in behavioral space.

Conclusion. The developed concepts support the idea of ontologization, identification, and problem-solving by an agent in a real-world environment as processes of autonomous behavior synthesis, which provides a theoretical foundation for the development of universal artificial intelligence systems.