Re: What is Computation?

From: HARNAD Stevan (
Date: Wed Jun 05 1996 - 18:57:55 BST

> From: "Nickson, Joanne" <>
> Subject: What is Computation?
> Computation is a process of symbol manipulation. It works according
> to a set of rules which operate using arbitrary symbols.

The symbols are arbitrary in that they do not resemble what they are
about, nor are they causally connected to it. The rules operate purely
on their SHAPES, not their meanings.

> The
> hardware (implementation) in computation is irrelevant becuse of the
> arbitrariness of the symbols.

Symbols with any "shape" will do.

> The important part of computation is
> the set of rules or algorithms which manipulate these symbols.
> Computations work on mechanical rules which operate purely upon the
> shape of the symbol without having any knowledge about what they
> actually mean. Algorithms, the mechanical rules, are mindless
> processes which means that computation needs no explanation like a
> homunculus. This eliminates the problem that imagery theory posed.

Correct, but it turns out analog processing doesn't need a homunculus

> As long as the correct procedure for manipulating the symbols is
> applied the outcome will be correct regardless of whether the process
> has been understood.

You are thinking of someone with a mind USING the symbol system (with or
without understanding), but remember that computationalists proposed
that what was going on in the mind may itself be symbol manipulation,
and for that it is no longer clear whether it is such an advantage that
the symbol manipulation is based on arbitrary shape rather than meaning:
The homunculus is eiminated, but what about meaning? This could be
related to Searle's Chinese Rooom Argument and the Symbol Gournidng

> An example of computation is the formula for a
> mathematical equation ie. y = a+bx. A computer (or the mind) need not
> understand what the symbols stand for, in fact it doesn't matter what
> the symbols are or what shape they take as long as the rules are
> learned and followed the result will be correct. The process of
> computation uses input which requires no understanding to produce
> correct output.

Well done. To get it over the top for an A, integrate it with some of
the bigger issues. (But this was a very high 2(1)!

This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:23:45 GMT