Models of Adaptive Behavior and Problem of Origin of Intelligence
Red'ko Vladimir G.
Scientific-Research Institute for System Studies, Russian Academy of Sciences, Moscow, 119333, Russia
vgredko@gmail.com
Abstract. An approach to the problem of evolutionary origin of intelligence is discussed; the approach is based on modeling of evolution of adaptive behavior. Works of leading laboratories in the field of simulation of adaptive behavior are characterized. A special attention is paid to reinforcement learning and adaptive critic designs. The project “Animat Brain” directed to development of a general platform for systematic designing of models of adaptive behavior is described. The results of investigation of the concrete model of evolution of self-learning agents that are based on adaptive critic designs are represented. The sketch program for future research of evolution of adaptive behavior is proposed.
Key words: adaptive behavior, problem of origin of intelligence, biological information processing systems, cognitive evolution, reinforcement learning.