Turing’s Approaches to Computability, Mathematical Reasoning and Intelligence

In this paper a distinction is made between Turing’s approach to computability, on the one hand, and his approach to mathematical reasoning and intelligence, on the other hand. Unlike Church’s approach to computability, which is top-down being based on the axiomatic method, Turing’s approach to computability is bottom-up, being based on an analysis of the actions of a human computer. It is argued that, for this reason, Turing’s approach to computability is convincing. On the other hand, his approach to mathematical reasoning and intelligence is not equally convincing, because it is based on the assumption that intelligent processes are basically mechanical processes, which however from time to time may require some decision by an external operator, based on intuition. This contrasts with the fact that intelligent processes can be better accounted for in rational terms, specifically, in terms of non-deductive inferences, rather than in term of inscrutable intuition.