Minds, Machines, and Turing
Abstract
In the 1960s, the philosophers J.R. Lucas and Paul Benacerraf presented
arguments against mechanism, the thesis that minds can be simulated by Turing
machines. Instead of discussing Turing machines directly, they shift their focus to
using Gödel's incompleteness theorems to show that theories of classical first-order
logic could not represent minds. They believed that these arguments about logical
theories implied that the mechanistic thesis for Turing machines was incorrect.
However, I show that their arguments are insufficient to topple the mechanistic
thesis for Turing machines. I then present a new argument, focused directly on
Turing machines, that seeks to show that mechanism is false. The argument is
reminiscent of the halting problem for Turing machines, and it proceeds by inference
to the best explanation to demonstrate that actual minds are not Turing-simulatable.