Marian Mazur, “Myślenie maszyn” (‘The Thinking of Machines’), the monthly Problemy, 1963, no. 9, pp. 535–545. The text is based on a lecture given in October 1962 in Ilmenau.
The answer this article gives to the question of machine thinking is that there is no reason to regard a thinking machine as impossible, but such a machine would act in its own interest. Designers who want a device that resembles a human being and at the same time serves human beings alone are, in the author's view, asking for something self-contradictory. Mazur reaches this conclusion by reversing the order of the debate of the time: its participants tried to settle whether a machine can think without having established what thinking is, whereas he begins with a definition.
The heart of the dispute comes down to a single organ. All three differences between a machine and an organism — how lasting the record is, how many correlational elements there are, and whether there is a homeostat — are, in the author's view, quantitative and do not mark a boundary that cannot be crossed, but it is the third that decides what kind of device one is dealing with. A technical regulator has every component of a control system except the homeostat, so it cannot tell beneficial stimuli from harmful ones; the criterion of usefulness is brought in by a human being, who changes its setting. However elaborate such a device may be, it remains an automaton; a homeostat would turn it into an autonom, controlling itself for the sake of its own safety.
The definition is built on a model of a system that processes information: receptors, a correlator with registrators and estimators, effectors, and the homeostat. The flow in the correlator is described by the correlational quantities: correlational power, correlational potential and correlational conductance, the last of which grows with every flow, which is how the author accounts for registering; the share of the homeostat is expressed by reflection potential. On this model Mazur explains emotion, reflection and decision, links thinking to the reflection circuit, and links consciousness to the connection between the correlator and the homeostat.
Among the articles discussed on this site, this is the first text in which the pair automaton–autonom appears; it will become the axis of Cybernetyczna teoria układów samodzielnych (‘A Cybernetic Theory of Autonomous Systems’, 1966). The introductory note presents the theory as a fragment of a book on character then in preparation; in Cybernetyka i charakter (‘Cybernetics and Character’, 1976) the definitions of thinking and consciousness already take a different form.