Re: Paper by ~MM on distributed self-awareness


Stephen Harris wrote:


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.


I don't think her paper means the same thing to me as it does to you.
Melanie has an integrated view. She describes her book with

"On one level, the work described here is about analogy-making, but on
another level it is about cognition in general, exploring such issues
as the nature of concepts and perception, and the way in which highly
flexible concepts emerge from a lower-level "subcognitive" substrate."

SH: I think this the same underlying process which "guides" the immune
system and from which self-wareness arises -- an emergent response.
Or, why do you think it is appropriate to describe the immune system
as "adaptive feedback mechanisms" which induce self-regulation." but
it is "simplistic" to describe the brain that way. I think the brain is part
of the body and it is the same mechanism of emergence which produces
diffuse consciousness, the immune response or fluid analogic thought.




For my part, my feelings are also that self-awareness and consciousness
for that matter are emergent phenomena of certain types of "complex"
systems, but to me it only makes sense to state things in a
__relative__ manner. IOW, you can't just say such systems as Melanie's
are "self-aware", instead you have to say something more like "they
exhibit some of the characteristics we attribute to self-aware 'living'
systems, but only to an extremely tiny extent". These programs are just
so simplistic that you have to make such overt qualifications, or it
just sounds silly.

I happen to have Hofstadter's 1995 book "Fluid concepts and creative
ananlogies" on the shelf, and notice that both Melanie and also
Chalmers were his students. This book also goes into significant detail
about CopyCat and introduces Metacat, so these ideas are well over 11
years old.

In the book H says on pg 309 section heading: "Copycat: self-aware, but
Very Little". I think that sums it up. Next page talks about
consciousness.

Basically, they're trying to make the case that "self-monitoring" is in
some way connected to "self-awareness", but this is really just the
same old quandary Searle talks about regards the difference between
doing and being. Just because something acts like it's conscious
doesn't mean it is conscious. Just because a program monitors it's
output and feeds back information [self-monitoring] doesn't make it
self-aware, rather it's just self-adaptive, AFAIAC. Etc, etc, etc.

As I said last time, H+M+C's are very LOOSE definitions, like saying a
thermostat is conscious. You just end up getting what you define. To
me, such arguments are just a waste of time, in fact. Good for
philosophy, maybe, but waste of time for engineering. As we've seen,
some people are happy to go on arguing for years and years about what
is and is not conscious, aware, etc. These are endless arguments, and
have no solution, and are a waste of time, AFAIAC.

OTOH, as I mentioned last time, where Melanie talks about creating
systems with architecture common to so-called "complex" systems, having
many interacting components, nonlinear feedbacks, self-organization,
patterned output, etc, there she's onto something you can actually get
your hands dirty with. IE, to produce something more than just words.




"adaptive feedback mechanisms" which induce self-regulation."
Tha phrase describes the function of analogy, which they think
is the basis of intelligence. A major factor in the evolution of
human intelligence is that it contributed to improved survival
responses which are interactively regulated by natural selection.
Analogy-Making as Perception by Melanie Mitchell .
(1) The perceptual process must be fine-grained, diverse, redundant,
and decentralized.

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.

Emergence is sometimes described as irreducible. Though the
output is caused by lower level entities, it cannot be predicted.
I think the human body containing the brain is a spectrum of
emergent behaviors which differ from each other not in their
underlying mechanism/process but are categorized subjectively
at "higher" levels of abstraction, more sophisticated behavior.

You think she muddies the water. I think she shows how the
body using the immune system for survival is similar to the body
using brain originated analogical responses to improve survival.
Both can be described as emergent behavior although both could
also be distinguished as different levels of emergent complexity.



I agree with this. As I noted where she muddies the water is trying to
make the analogy between ants/immunue self-awareness and human
self-awareness. If she were to qualify her statements, as I mentioned
above, she'd avoid a lot of useless contention.



Melanie Mitchell: "Implications for Artificial Intelligence ...

"The four principles listed above, along with other general
principles abstracted from the study of decentralized complex
adaptive systems, can be a guide in designing articial intelligence
systems with decentralized architectures that have sophisticated
abilities for pattern perception and self-awareness.

In fact, these principles guided the design of the Copycat ***
system, developed by Douglas Hofstadter and myself (Hofstadter
& Mitchell 1994; Mitchell 2001), and its successor, Metacat,
developed by Hofstadter and James Marshall (Marshall
& Hofstadter 1998; Marshall 2002)."

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

Well the paper is linked to the immune system -->Copycat ***
"In fact, these principles guided the design of the Copycat"
and
Analogy-Making as Perception by Melanie Mitchell .
"In this book, Melanie Mitchell describes Copycat,

Book:
"On one level, the work described here is about analogy-making, but
on another level it is about cognition in general, exploring such issues
as the nature of concepts and perception, and the way in which highly
flexible concepts emerge from a lower-level "subcognitive" substrate."

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

SH: IMO, 'the highly flexible concepts which emerge from the
lower-level "subcognitive" substrate' have the properties described
in (1) 'fine-grained, diverse, redundant and decentralized.

The brain is redundant and if one part of the brain is damaged
there is usually a portion of the brain which can serve as backup.



Yes, as I mentioned these are the key features of so-called complex
systems.


I think Melanie brings to light a relationship that might not be
understood. There is a continuum of causal but unpredictable
emergent systems which serve each other as links in a chain.
If your focus was on a Jim Brommer perspective of what
constitutes analogical reasoning, then you have a level of
intelligence which experiences itself as self-awareness and
diffuse consciousness. Portions of the brain can be removed
and the individual still considers themself as an I, self-aware.

But if your perspective shifts to a molecular biology analysis,
you will use words which reflect your bias in contrast to an
evolutionary (social) psychologist. I meant to an observer
it would be difficult to distinguish the scope or range of
potential controlled behaviors just looking at the external
behavior of an ant foraging or taxi cabs looking for fares.
I could have said, just judging from appearances, which
is smarter, a group of chimpanzees building a sleeping nest
from branches, or a six foot high termite mound when
you judge from the complexity of the structures.



Actually, the basic concepts of complexity theory I menntioned last
time pertain to systems ranging from fluid mechanics all the wway to
galaxies, and lots inbetween.


Did you
read John McCarthy's claim that a thermostat feels cold?



As I indicated, I don't much care to attribute feelings to thermostats.
You get what you define.

The only statement regards thermostat "consciousness" that makes any
sense to me is that .... "IF we define consciousness such that a
thermostat can be conscious, THEN it's level of consciousness is many
many orders of magnitude less than that of a human, but any reasonable
system of measurement". Or some such. Likewise for its ability to
"feel".



I think from a systems approach, an AI is intelligent when
it answers email that doesn't make the human reading it
wonder if the sender is a weirdo. That is a judgment based
on observing external behavior. One thinks another is
intelligent from their behavior and also maybe if they look
like you they can think like you. But if you want to find
the potential range of behavior you look at the interior
mechanism which regulates behavior. It may be more
limited than what you infer the organism is capable of
just from a sampled series of external behaviors; or it
may be more create/expanded than what you might infer
or predict from a sampled series of external behaviors.
Perhaps emergence describes this latter case. AI is not
going to progress very far if intelligence is emergent and
the substrate cannot be controlled to produce directed
evolutions of predictable hierarchal emergent structures.



Yes, so a reasonable approach might be to first study what general
attributes are present in various examples of other complex systems
from nature. There appear to be some general rules for how these are
composed and how they operate. Each is different in its own way, but
all seem to follow similar general guidelines.



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.



As I mentioned, 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....".

The next sentence is "... We expect that near a bifurcation,
fluctuations or random elements would play an important role, while
between bifurcations the deterministic aspects would become dominant".
Melanie's describing a classical complex system, ala complexity theory.



Perhaps the water looks muddy because the solution is
not actually clear, or at least to Melanie, it appears fuzzy.

The elephant's memory held two tales,
Stephen

.

Relevant Pages

  • Re: Does Searles "Chinese Room" argument imply that consciousness is non-scientific?
    ... emergent property of interacting items. ... Don't tell me you're doing science when you're doing ... I assert that consciousness is physical. ... All it requires is to stick probes into your brain and have you ...
    (comp.ai.philosophy)
  • Re: origin of thoughts
    ... > brain is aware of what another part of the brain is doing, ... One of the things that is said to "show" consciousness, ... sense of self-awareness, is that people may report on their behavior both ... > to have the actual infinite progression and that this would ...
    (sci.bio.evolution)
  • Re: Is sentience an emergent brain behavior?
    ... to emergence as an "explanation"for consciousness. ... I can draw lines, and my brain ... It is still an emergent behavior. ... If a rock rolls down a hill and stops, ...
    (comp.ai.philosophy)
  • Is sentience an emergent brain behavior?
    ... to emergence as an "explanation"for consciousness. ... I can draw lines, and my brain ... It is still an emergent behavior. ... If a rock rolls down a hill and stops, ...
    (comp.ai.philosophy)
  • Re: Is sentience an emergent brain behavior?
    ... to emergence as an "explanation"for consciousness. ... I can draw lines, and my brain ... It is still an emergent behavior. ... If a rock rolls down a hill and stops, ...
    (comp.ai.philosophy)