Learning to control intelligently

The author attempts firstly to give his own perception of what intelligent control is, by identifying both the characteristics of those problems which can arguably only be solved satisfactorily by a control system which possesses some degree of intelligence, and the consequent characteristics of such an intelligent control system. He then proposes an architecture for an intelligent control system which is based on these observations and which is also inspired by the recent results of computational and experimental neuroscience. In his opinion the use of neuroscience as a basic inspiration for the future development of intelligent control systems is essential.