From: David Yeo 
Organization: Ontario Institute for Studies in Education, Toronto
Date: Sun, 17 Nov 1996 13:22:25 GMT

On 16 Nov 1996, Anders N Weinstein wrote:
 
> Atoms do not compute anything, since they don't have representations. 
> Their behavior might (or might not) be described by deterministic laws, but
> again that is irrelevant, or else you could just explain human behavior
> at the physical level and claim to have a cognitive theory. 

In a sense don't atoms compute the charges which act upon them?  

I suggest that your claim that representation is necessary for computation
critically hinges on whether the term "representation" is viewed as an
interpreted abstraction which substitutes for the "real world" or merely
a state reflecting some subset of it.  The former implies an intelligent
abstracting agent, the latter (generally) does not. 

  Note: Before someone points out that computers have representations,
        remember computer representations are a function of programme,
        i.e.  human, design.

Moreover, your comment would seem to imply that it is possible to isolate
structure (representation) from function (computation).  While I realise
that this is a key tenet of the functionalist paradigm (a la David Marr),
IMHO the claimed independence of structure and function is largely false
(at least in the context of complex systems).  For in complex systems the
interaction between structure and function predominates, i.e. structure
tightly determines function, and function modifies structure.  In other
words, structure and function are merely two sides of the same coin. 

Cheers,

- David Yeo (Applied Cognitive Science, University of Toronto)