The many accounts of digital computation “establish different (but not all irreducibly different) requirements for a physical system.” This paper addresses the physical meaning of digital computation. Instead of offering a new account of concrete computation, it gives two rather straightforward arguments: concrete digital computations entail nontrivial, distinct physical requirements, and computational frontiers are usually linked to Turing’s abstract formalism of computability, without concrete specifications. Surprisingly, philosophers and cognitive scientists ignore these arguments. The author states that “[he] shall remain neutral on whether cognition can indeed be fully explained computationally.”
For some reason, the paper does not compellingly exercise the so-called Church-Turing thesis. This may be due to its insufficient mathematical rigor, lending the unanswered computation-cognition question to engineering rather than a philosophical treatment. Informally, the Church-Turing thesis states that all reasonable computational models are algorithmically equivalent, differing only in performance. Apparently, cognition must rely on content-addressable memory with some intelligent resolution of multiple responses, such as in Google PageRank. The most suitable realization of such a scheme is through cloud computing.