... computation".¹

One ambiguity that is present, but doesn't seem to be the source of the confusion I am addressing, we might call ``Clinton's qualifier": that the meaning of the claim ``cognition is computation" depends on what the meaning of ``is" is. The copula can be taken to indicate various degrees of metaphysical commitment. It might stand for the symmetrical relations ``is intensionally identical with" or ``is extensionally identical with". Or it could stand for an asymmetrical relation, yielding either 1) cognition reduces to computation, or 2) one can (must?) use computational concepts in order to distinguish cognitive from non-cognitive systems, or 3) cognition is best explained using computational concepts, etc. I hope that the points I make in what follows apply no matter which of these readings one prefers.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... phenomena).²

Although this distinction between the intuitively computational and theoretical attempts to account for such may now be thought to be radical, it was at the heart of original theoretical thinking in the field. Church's and Turing's theses were that their respective formalisms did justice to a prior, intuitive notion of computation and the computable. It is the fact that these theses bridge the theoretical and the intuitive that makes them unprovable in any formal sense.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... water).³

This example suggests, therefore, that what we take to be computational not only depends on our theory of computation, but on our other theories as well. We exclude whales from the class of fish because of the explanatory advantages of seeing them as mammals as much as the explanatory drawbacks of seeing them as fish. So also might we deny computational status to an information-processing device if we have a non-computational theory which provides better explanations of it.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... intentionality?⁴

Compare the inertial Watt governor with the super sunflower in [14].

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... nature⁵

For example, perhaps one could arrange the dynamics of the governor such that there was a not only a first order correspondence between governor and engine speed during disconnection, but a second order correspondence as well. I.e., consider a Watt governor that, if it was detached from the engine during a period of engine speed increase, would increase its own speed during the disconnection, at an appropriate rate.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... question⁶

Why not? Because any Turing machine which supposedly could would appear as some k=K in the list of Turing machines. When it got around to considering the case of K, it would have to halt if it doesn't halt. But that's impossible, so it doesn't halt. Yet it cannot halt to indicate that fact. See Penrose [8] for a clear explanation.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... can.⁷

The main reason for believing this is we humans can follow the reasoning in the previous footnote.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... conclusions.⁸

I believe this kind of transparent computationalist move was originally proposed by Putnam.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.