Distributed Adaptive Control: A Proposal on the Neuronal Organization of Adaptive Goal Oriented Behavior

摘要: In behavioral motor coordination and interaction it is a fundamental challenge how an agent can learn to perceive act in unknown dynamic environments. At present, not clear – without any explicitly predefined knowledge acquire internal representations of the world while interacting with environment. To meet this challenge, we propose biologically based cognitive architecture called Distributed Adaptive Control (DAC). DAC organized three different, tightly coupled, layers control: reactive, adaptive contextual. systems are self-contained fully grounded, meaning that they autonomously generate their primary sensory inputs, hence bootstrapping behavior form simple advance interactions. Following approach, have previously identified novel environmentally mediated feedback loop organization perception behavior, i.e. feedback. Additionally, could demonstrated dynamics memory structure DAC, acquired during foraging task, equivalent Bayesian description foraging. chapter present concise show allowing us extend different subsystems more biophysical detailed models. These further developments architecture, only allow better understand biological systems, but moreover DACs capabilities generality.