Ethological dynamics in diorama environments

Biological systems possess the capability and intelligence to adapt to their environment. While the information processing by higher organisms with brains is monolithic, the adaptability and intelligence of unicellular organisms or cell populations are distinctly decentralized. The fundamental principles characterizing the optimality of such decentralized information processing remain unclear. This research aims to extend existing learning and control theories that presume the centralized intelligence. By utilizing mean-field control and mean-field games as a basis, we construct a new theory of biological information processing capable of handling decentralized information processing and control. By deriving from this theory a cellular dynamics that realizes optimal decentralized information processing, we explore the optimality of proto-intelligence. This theory, linking the optimality in decentralized information processing to cellular dynamics, will contribute as a foundational theory of proto-intelligence.