🚀 go-pugleaf

In article <34307e12.30875895@news.earthlink.net>
           savainl@earthlink.net "Louis Savain" writes:

> In article <875446625snz@longley.demon.co.uk>,
> David@longley.demon.co.uk (David Longley) wrote:
> 
> >[...]
> >Elsewhere I have asked (rhetorically) whether it really is AI  to 
> >build and implement models of classical conditioning as  Rescorla 
> >and  Wagner did the late 60s and early 70s (and which Sutton  and 
> >Barto  contributed to later in the 80s). There is, in my  view  a 
> >considerable  difference  between AI (which may cash  out  to  be 
> >science  and technology in general) and  experimental  psychology 
> >and the neurobiology of learning. Those interested should look at 
> >Fragjn97.pdf at the www below.
> 
>   It seems that current research in AI is mostly involved with finding
> new and clever programming methods to do fancy things.  In this light
> I agree that AI has little to do with natural intelligence.  It is for
> this reason that the relatively new field of computational
> neuroscience was born, as a way for a group of researchers to distance
> themselves from formal logic, knowledge representation, symbol
> manipulation, frames, scripts, etc., and even ANNs.  However, the
> original goal of AI (to create a synthetic intelligence whose behavior
> and learning prowess would be commensurate with that of humans) is
> still alive and kicking, IMO.  That AI research went astray right from
> the start is no reason to abandon the early spirit.  If anything, I
> think most of the current work in AI should be renamed to "complex
> logical solutions" or something similar, so as to retain the name AI
> for the work in computational neuroscience.  But then again, it may
> not matter all that much.
>   As an aside, my own limited understanding of this subject leads me
> to conclude that intelligence, even though highly distributed within
> constrained areas, is indeed fragmented.  I see a family of
> cooperating subnetworks or agents, if you will, that nevertheless
> remain ignorant of each other's operation in significant ways.  Even
> within the individual agents themselves, I see a large degree of
> fragmentation which would become apparent were there to be a local
> malfunction of some sort.  This, I understand to be your own view of
> human intelligence.


Yes.

What's now being called computational neuroscience is really just 
that part of learning theory which is anchored or constrained not 
only by the empirical animal data, but also by observed  function 
of  presumed neural substrates. Most of that informative work  is 
still  done  at a low level, ie in simple organisms or  in  small 
subsystems of animals. 

The sort of "decision making" made by your average human being or 
lab  mammal is still best modelled by conditioning -  which  from 
another   perspective  is  folk  psychology  with  all   of   its 
deficiencies.

What  I  want to draw attention to is the fact that even  if  the 
work in ANN research proved successful, ie that such systems were 
engineered,  the results would be what we would generally  regard 
as superstitious behaviour. 

To produce intelligent behaviour one needs to inculcate extensive 
skills which are situationally apposite, and I think those skills 
will  turn  out to be procedural but embedded  -  which  probably 
means, not as intelligent as we imagine <g>. That's what I  meant 
when  I  said that AI just *is* science - perhaps I  should  have 
said, our scientific knowledge base. 

I  once  thought that the CYC project had grasped that -  but  it 
hasn't  -  it's in pursuit of the original  McCarthy-Hayes  -goal 
which  I  think is misguided to the extent that 1)  it  seeks  to 
include folk notions and 2) is practically undeliverable even  if 
it didn't. 

What these folk are getting close to realising is that there is a 
need for good scientific literature databases, improved  versions 
of INDEX MEDICUS, PSYCHLIT etc. 

I wish you luck in your R&D though.

-- 
David Longley (check end reply line #)

Longley Consulting                                                  London, UK
Behaviour Assessment & Profiling Technology,
Research, Data Analysis and Training Services,
Small IT Systems                                http://www.longley.demon.co.uk