Re: cognitive != observable?

In message 
<[EMAIL PROTECTED]>, 
Anthony Bucci <[EMAIL PROTECTED]> writes
> > >>>Secondly, what is widely referred to as 'cognitive behaviour', invariably
> > >>>reduces to publicly observable and measurable behaviours - a fact which
> > >>>renders the qualifier 'cognitive' redundant except as a convenient
> > >>>sub-classifier.
>
> That above is a thesis, not a fact.  Indeed, it's an important thesis
> which has been batted around for awhile but is still not settled.  The
> most interesting alternative to me deals with the idea that a non-trivial
> high-level behavior can "emerge" out of the simple interactions of
> low-level behaviors.  A classic example is water: hydrogen and oxygen are
> both gases, and a simple-minded addition of them would lead to another
> gas, but instead it leads to water (and a big explosion if you're not
> careful).  Likewise, we might know what individual neurons do, but a big
> interconnected network of them can do other things like prove theorems or
> argue on USENET which are not obviously predictable from the actions of
> the individual neurons.
>
> A converse to that: if we can observe the high-level behaviors of a
> system, we cannot necessarily predict which low-level component behaviors
> led to the emergence of those high-level behaviors.  In other words, the
> cognitive needn't be reducible to the observable behaviors, because that
> amounts to saying that by watching the high-level behavior of an organism
> you can predict what the various parts of its brain were doing.  We can't
> do that now, and it's possible that in principle it can never be done
> fully.
>
> Surely this is one of the most important unanswered questions in science
> right now.
Hi,

I do understand what you are saying, and I *will* look into what you've 
mentioned, but what I'm suggesting here and elsewhere is quite radical 
so bear with me when I suggest what I do below:

I'm going to ask you to look into the context of where that extract came 
from, what the system was endeavouring to do, what the papers which 
supplement it on "What Works" review (at the same website) and what this 
suggests. This will explain *why* I made the statement that I did in 
that extract.

having done that, you might, I hope, reconsider and appreciate better 
what I'm saying, and why I'm saying it.

Whatever we do, be it laboratory based or in a wider applied context, we 
are always working with behaviours - how we *talk* about this may change 
a little, but in the final analysis, we work with behaviour. It is 
easier *not* to see that when you are working with a limited set of that 
behaviour (usually verbal behaviour) in an academic context. It is much 
*harder* not to see this when one works in an applied context where 
outcome or change is a Key Performance Indicator. In this particular 
situation, I think the applied psychologist has something useful to say 
to the academic psychologist. In these contexts, one *could* say that 
"cognitive" is either a metaphor for a complex of behaviours, or it 
refers rather explicitly to a collection of skills which can loosely be 
called "verbal" behaviours. Bear with me through this:


http://www.longley.demon.co.uk/Frag.htm

or

http://www.longley.demon.co.uk/Fragjn97.pdf


and for the application area:

What Works & What Can be Effectively Managed: A Close Look at the Data 
(1997a) http://www.longley.demon.co.uk/Sm-97apr.pdf

A Further Comment on Recent Claims in the 'Rehabilitation of
Rehabilitation Literature..(1997b)
http://www.longley.demon.co.uk/Workj97.pdf

and some relevant recent research.

http://www.homeoffice.gov.uk/rds/pdfs2/r206.pdf
http://www.homeoffice.gov.uk/rds/pdfs2/r205.pdf

> These ideas have been formalized in an interesting way by A. Ehresmann (a
> mathematician) and J.P. Vanbremeersch (a physician) in the form of "memory
> evolutive systems."  They are attempting to capture this idea of
> non-trivial behaviors emerging out of the interactions of low-level
> components.  One point they demonstrate is exactly how the interconnection
> of the components can lead to higher level behaviors which are not cleanly
> predictable unless you know those interconnections.  Clearly there are
> important implications of that observation, since we can rarely know all
> the interconnections which exist among a system's components --
> particularly when the system is as complicated as a brain.  Others in the
> artificial life community, for instance Nils Baas, do similar work.
> There is a forthcoming special issue of the journal Artificial Life on
> Dynamical Hierarchies which is all about this question.
>
> Anthony


--
David Longley