H. ATLAN states as follows the problem of comparative reliability of natural and artificial machines:
"A reliability like the one of the brain, able to function continuously, while cells die every day without replacement, with unforeseen changes in the blood 's flow, volume and pressure fluctuations, not to mention amputation of important parts, which perturbs only in a very limited fashion the whole set's performance, is not matched in any artificial automata. This fact had already struck J.von NEUMANN (1966), who was seeking to better the computer's reliability and could not imagine such a reactive difference to random environmental factors of aggression (noise), if not related to a basic difference in the organizational logic of the system. "Organisms, with their faculty to "swallow" noise, could not be understood as just only somewhat more reliable than the known artificial machines, but as systems whose reliability could be explained only by qualitatively different principles of organization" (1972, p.23)
"VON NEUMANN, WINOGRAD and COWAN's (COWAN, 1965) investigations aiming at the discovery of construction principles for automata whose reliability would be greater than their components'…led to the definition of necessary (and sufficient) conditions of feasibility of such automata). Most of these conditions, (redundancy of components, redundancy of functions, complexity of components, delocalization of functions) lead to a kind of compromise between determinism and indeterminism in the build up of the automata, as if some measure of indeterminacy should be necessary, from a certain level of complexity on, to allow the system to adapt itself to a certain level of noise" (ATLAN, 1972, p.23).