Outline of a theory of thought-processes and thinking machines

Abstract

Abstract Thought-processes and certain typical mental phenomena are schematized into exact mathematical definitions, in terms of a theory which, with the assumption that learning is a relatively slow process, reduces to two sets of equations: “neuronic equations”, with fixed coefficients, which determine the instantaneous behavior, “mnemonic equations”, which determine the long-term behavior of a “model of the brain” or “thinking machine”. A qualitative but rigorous discussion shows that this machine exhibits, as a necessary consequence of the theory, many properties that are typical of the living brain: including need to “sleep”, ability spontaneously to form new ideas (patterns) which associate old ones, self-organization towards more reliable operation, and many others. Future works will deal with the quantitative solution of these equations and with concrete problems of construction—things that appear reasonably feasible. With a transposition of names, this theory could be applied to many sorts of social or, more generally, “collective” problems.