Re: Paper by ~MM on distributed self-awareness


Stephen Harris wrote:
"feedbackdroids" <feedbackdroids@xxxxxxxxx> wrote in message
news:1146251861.797483.138130@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

Stephen Harris wrote:
http://www.cs.pdx.edu/~mm/self-awareness.pdf
by Melanie Mitchell (Hofstadter was her Phd. supervisor)



I had a chance to read the rest of the article now, and Melanie is
basically talking about modeling AI's as emergent dynamical systems ala
the application of the ideas of complexity theory. From pg 5 ....

============
However, the philosophy behind the programs can be summarized by the
following principles for modeling perception, which closely follow the
principles abstracted above.

(1) The perceptual process must be fine-grained, diverse, redundant,
and decentralized.

(2) Perception is guided by "fluid" concepts, which are themselves
shaped as the perceptual process unfolds.

(3) The perceptual process proceeds as an interplay of bottom-up modes
(driven by stimuli from the environment) and top-down modes (driven by
expectation, prior knowledge, biases, and what has already been
discovered). This interplay is not preprogrammed, but is an emergent
effect of the collective actions of low-level components of the system.

(4) The perceptual process shifts over time from being highly parallel,
random, and bottom-up, to being more focused, deterministic, and
top-down. As in (3), this shift is not pre-programmed, but rather is an
emergent effect of collective behavior in the system.

(5) The perceptual process must have a means of "self-watching"
monitoring its own state and progress that feeds back to affect
behavior.
==================

Compare this to complexity theory ideas that self-organization,
patterning, and phase shifts occur in systems that have: (a) many
simple interacting components, (b) nonlinearities, (c) positive and
negative feedback, and (d) continual energy flows which produce (e)
dissipative structures. Such systems have many self-organized operating
modes or attractor basins, multiple bifurcations at critical points
which make future behavior difficult to predict, and global [top level]
effects which feed back and down to constrain behavior of bottom-level
elements.

Compare #(4) especially to Prigogine's statement that ... "...
self-organization processes in far-from-equilibrium conditions
correspond to a delicate interplay between chance and necessity,
between fluctuations and deterministic laws ...".

Melanie is basically giving an example of how to apply this.


-------------------------------------------------------------

Analogy-Making as Perception by Melanie Mitchell .
"In this book, Melanie Mitchell describes Copycat, a computer model
of analogy-making, developed by Douglas Hofstadter and herself. This
work is based on the premise that analogy-making is fundamentally a
high-level perceptual process, in which perception, interacting with
concepts, gives rise to "conceptual slippages" which allow an analogy
to be made. Copycat is a model of this complex, subconscious interaction
between perception and concepts that underlies the creation of analogies.

In Copycat, both concepts and high-level perception are emergent
phenomena, arising from large numbers of low-level, parallel,
non-deterministic activities. In the spectrum of cognitive-modeling
approaches, Copycat occupies a unique intermediate position between
symbolic and connectionist systems, a position that is at present the
most useful one for understanding the fluidity of concepts and
high-level perception."

-----------------------------------------------------------------
feedbackdroids wrote:
"Unfortunately, I think Melanie seriously muddies the water in the first
half of the paper, by trying to call immune systems and ant colonies
"self-aware", and somehow similar to the brain. She admits that this is
a contentious declaration, and that many would say the systems
described are simply "adaptive feedback mechanisms" which induce
self-regulation."

SH: I think she means the foundational substrate of analogy making
is at the cellular level of the brain and not described well as a process
of the intellect at the category fitted or matched to self-awareness.
Analogy making is perceptual rather than rational mentation. But
if you look towards the surface or penthouse then analogy making
seems like an intellectual endeavor fitting self-awareness/purpose.



Ooof, you've patched several different things together here.

My comment was regards her paper, and now you've patched in
something about her book above it, which wasn't in your first post.
So I have no idea what your new post is really about. Makes things
even muddier. I didn't notice anything about analogy making in the
paper, other than a passing comment regards something not talked
in the paper.



The symbolic AI thesis looked at behavior. The Turing Test
measured an external criteria provided by subjective humanity.
The output was the sole determinant, not how it was produced.

The ant colony behaves in an organized way. The activity of ants
looks a lot like a fleet of Yellow cab taxis darting around New York.
But the ants have no organizing mind while the taxis have a dispatcher
which controls the paths of the cabs. So the evaluation of the internal
or intrinsic mechanism which drives external behavior is seen to
provide an added predictive power which is unavailable and not
foreseen by the eventualities/boundaries acquired by looking at
behavior alone.
One can say all behavior stems from the cells/neuron of the brain but
a metalevel of description is necessary to describe how consciousness
arises, the sense of self-awareness and the experience of Iness.
I think the choice of terms depends upon the level/scope of foucs.



???????? does this have anything to do with her paper? She doesn't
really talk about the brain vis-avis consciousness to any extent.



Because an analogy can be made that an ant is like a taxi cab
because they both forage the environenment and we know that both
are composed of atoms which are organized into more complex
molecular structures having unique properties says how they are
alike but doesn't outweigh the preponderance of differences into an
apt/precise analogy. I write this while I am tired. It boils down to I
think self-awareness emanates as the tip of a perceptual iceberg.



OK, I guess you were responding to something other than her paper
and what I wrote about it.

Again, she was talking about entities much more simplistic than the
human brain, namely ants and the immune system, and indicating they
have "self-awareness" [or a similar mechanism as such] somehow
comparable to the brain. What they really is some semblance of
self-organizing capability, which results in patterned activity. This
is hardly self-awareness, except as only very loosely defined.
Bascially, output monitoring and feedback are not self-awareness
as we think of such regards the brain. A simple radio automatic
gain circuit woiuld be self-aware using this definition..

However, I seriously doubt that writing programs like Copycat
will make self-aware AI's, except by the loosest of definitions.
If you follow this line of reasoning, then Prigogine's thermodynamics
systems should also be self-aware.

I wish you had addressed the paper, rather than what wasn't in
the paper.


Lettuce and roots,
Stephen

.

Relevant Pages

  • Re: Paper by ~MM on distributed self-awareness
    ... Perception is guided by "fluid" concepts, ... Analogy-Making as Perception by Melanie Mitchell. ... and somehow similar to the brain. ... I think she means the foundational substrate of analogy making ...
    (comp.ai.philosophy)
  • Re: Paper by ~MM on distributed self-awareness
    ... The perceptual process must be fine-grained, diverse, redundant, ... Perception is guided by "fluid" concepts, ... Analogy-Making as Perception by Melanie Mitchell. ... I think she means the foundational substrate of analogy making ...
    (comp.ai.philosophy)
  • Re: OOP/OOD Philosophy
    ... > between a model created by analogy, and a model created by analysis? ... common with our perception of the world. ... > structures in the problem domain. ... I see it as a central tool. ...
    (comp.object)