Architecture of cognitive systems and their plasticity

The current large-scale transition to intelligent and robotic technical systems necessitates the understanding and development of corresponding design technologies that ensure this shift is implemented systematically. Special attention is required for systems possessing cognitive functions that enable learning and adaptive behavior based on experience gained through interaction with a dynamic environment. In this work, the authors aim to clarify the conceptual meaning of a cognitive system or cognitive automaton (cognimat) and explore their possible architectural designs, which represent the structural framework and composition of these next-generation systems. A separate issue involves defining an appropriate theoretical foundation that provides a representation of the architecture and properties of the cognitive domain, as well as the microarchitecture (or internal organization) of agents and cognitive tools that serve as key components of this domain. The primary requirements for cognitive systems or cognimats include: goal-directed behavior (inherent to all systems by definition), selfregulation and adaptive behavior based on experience acquired through interaction with a dynamic environment (self-learning). The third objective addressed in this work is identifying parallels between the architectural structure of the cognitive sphere in next-generation systems (artificial systems) and the organization and functioning of the nervous systems of mammals and humans (natural systems). Such parallels are intended to reveal commonalities between living and artificial systems, which could significantly influence the development of perspectives and theoretical foundations essential for understanding the organization and operation of these systems.