The following might be of interest to people interested in building
embodied AIs.

I found a reference in the following cog.sci paper to Tani's robotic
work, mentioned below. The approach involves recurrent NN's and
self-learning schemes, integration of bottom-up perception and
top-down "subjective" prediction processes, in a real robot in a
real-world environment. The philosophical citation is just for
reference. Tani's dynamical systems approach is the meat. This
approach attempts to obviate the need for the homunculus.

http://www2.canisius.edu/~gallaghr/tics2000.html

"Philosophical Conceptions of the Self: Implications for Cognitive
Science", Shaun Gallagher, 2000.

"... Tani (1998) explores the possibility of establishing an
artificial version of Strawson's minimal self in a machine ... Tani,
however, in contrast to Strawson, makes it clear that the robotic self
he is designing is the result of physical interaction between the
robotic body and its environment...."

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

http://www.bdc.brain.riken.go.jp/~tani/
http://citeseer.nj.nec.com/tani98interpretation.html

"dan michaels" <dan@oricomtech.com> wrote in message
news:4b4b6093.0402111226.7d496014@posting.google.com...
The following might be of interest to people interested in building
embodied AIs.

I found a reference in the following cog.sci paper to Tani's robotic
work, mentioned below. The approach involves recurrent NN's and
self-learning schemes, integration of bottom-up perception and
top-down "subjective" prediction processes, in a real robot in a
real-world environment. The philosophical citation is just for
reference. Tani's dynamical systems approach is the meat. This
approach attempts to obviate the need for the homunculus.

http://www2.canisius.edu/~gallaghr/tics2000.html

"Philosophical Conceptions of the Self: Implications for Cognitive
Science", Shaun Gallagher, 2000.

"... Tani (1998) explores the possibility of establishing an
artificial version of Strawson's minimal self in a machine ... Tani,
however, in contrast to Strawson, makes it clear that the robotic self
he is designing is the result of physical interaction between the
robotic body and its environment...."

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

http://www.bdc.brain.riken.go.jp/~tani/
http://citeseer.nj.nec.com/tani98interpretation.html

[snip]

I become very leery eyed when I read this kind of stuff.

Bold unsubstantiated statements such as,

"Self consciousness emerges when the relation between
the subject (top down process) and the object (bottom
down process) becomes incoherent."

Oh really? How would you verify that? I would find it
more convincing if you could say that "when a system,
capable of speech, becomes incoherent it will declare
itself as being self conscious as a result of ...."

I don't see how ad hoc definitions help us understand
these issues. To define the "subjective mind" (what other
kind is there?) as a "top down interpretive process"
seems kind of superfluous? Why not just call it TDIP?
Why pretend you have explained anything? It is not
required to understand how the machine works or
what is meant physically by incoherence.

How best to integrate higher level processing with the lower
level processing does not require "mind" or "consciousness"
to be part of the equation, except as an emergent property
(a prediction of this kind of system) which would be a major
breakthrough in our understanding of what those terms mean
in a physical sense.

Not all that glitters is gold and my gut feeling is they have not
hit the gold yet.

--
John Casey

"dan michaels" <dan@oricomtech.com> wrote in message
news:4b4b6093.0402111226.7d496014@posting.google.com...
The following might be of interest to people interested in building
embodied AIs.

I found a reference in the following cog.sci paper to Tani's robotic
work, mentioned below. The approach involves recurrent NN's and
self-learning schemes, integration of bottom-up perception and
top-down "subjective" prediction processes, in a real robot in a
real-world environment. The philosophical citation is just for
reference. Tani's dynamical systems approach is the meat. This
approach attempts to obviate the need for the homunculus.

http://www2.canisius.edu/~gallaghr/tics2000.html

"Philosophical Conceptions of the Self: Implications for Cognitive
Science", Shaun Gallagher, 2000.

"... Tani (1998) explores the possibility of establishing an
artificial version of Strawson's minimal self in a machine ... Tani,
however, in contrast to Strawson, makes it clear that the robotic self
he is designing is the result of physical interaction between the
robotic body and its environment...."

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

http://www.bdc.brain.riken.go.jp/~tani/
http://citeseer.nj.nec.com/tani98interpretation.html

[snip]

I become very leery eyed when I read this kind of stuff.

Bold unsubstantiated statements such as,

"Self consciousness emerges when the relation between
the subject (top down process) and the object (bottom
down process) becomes incoherent."

.............
How best to integrate higher level processing with the lower
level processing does not require "mind" or "consciousness"
to be part of the equation, except as an emergent property
(a prediction of this kind of system) which would be a major
breakthrough in our understanding of what those terms mean
in a physical sense.

Not all that glitters is gold and my gut feeling is they have not
hit the gold yet.

Yeah but it SOUNDS smart. Thus, Dan is impressed.

"JGCasey" <kjcasey@hotkey.net.au> wrote in message
news:402af50d_1@news.iprimus.com.au...

"dan michaels" <dan@oricomtech.com> wrote in message
news:4b4b6093.0402111226.7d496014@posting.google.com...
The following might be of interest to people interested in building
embodied AIs.

I found a reference in the following cog.sci paper to Tani's robotic
work, mentioned below. The approach involves recurrent NN's and
self-learning schemes, integration of bottom-up perception and
top-down "subjective" prediction processes, in a real robot in a
real-world environment. The philosophical citation is just for
reference. Tani's dynamical systems approach is the meat. This
approach attempts to obviate the need for the homunculus.

http://www2.canisius.edu/~gallaghr/tics2000.html

"Philosophical Conceptions of the Self: Implications for Cognitive
Science", Shaun Gallagher, 2000.

"... Tani (1998) explores the possibility of establishing an
artificial version of Strawson's minimal self in a machine ... Tani,
however, in contrast to Strawson, makes it clear that the robotic self
he is designing is the result of physical interaction between the
robotic body and its environment...."

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

http://www.bdc.brain.riken.go.jp/~tani/
http://citeseer.nj.nec.com/tani98interpretation.html

[snip]

I become very leery eyed when I read this kind of stuff.

Bold unsubstantiated statements such as,

"Self consciousness emerges when the relation between
the subject (top down process) and the object (bottom
down process) becomes incoherent."

Oh really? How would you verify that? I would find it
more convincing if you could say that "when a system,
capable of speech, becomes incoherent it will declare
itself as being self conscious as a result of ...."

I don't see how ad hoc definitions help us understand
these issues. To define the "subjective mind" (what other
kind is there?) as a "top down interpretive process"
seems kind of superfluous? Why not just call it TDIP?
Why pretend you have explained anything? It is not
required to understand how the machine works or
what is meant physically by incoherence.

How best to integrate higher level processing with the lower
level processing does not require "mind" or "consciousness"
to be part of the equation, except as an emergent property
(a prediction of this kind of system) which would be a major
breakthrough in our understanding of what those terms mean
in a physical sense.

Not all that glitters is gold and my gut feeling is they have not
hit the gold yet.

--
John Casey

"JGCasey" <kjcasey@hotkey.net.au> wrote in message
news:<402af50d_1@news.iprimus.com.au>...
"dan michaels" <dan@oricomtech.com> wrote in message
news:4b4b6093.0402111226.7d496014@posting.google.com...
The following might be of interest to people interested in building
embodied AIs.

I found a reference in the following cog.sci paper to Tani's robotic
work, mentioned below. The approach involves recurrent NN's and
self-learning schemes, integration of bottom-up perception and
top-down "subjective" prediction processes, in a real robot in a
real-world environment. The philosophical citation is just for
reference. Tani's dynamical systems approach is the meat. This
approach attempts to obviate the need for the homunculus.

http://www2.canisius.edu/~gallaghr/tics2000.html

"Philosophical Conceptions of the Self: Implications for Cognitive
Science", Shaun Gallagher, 2000.

"... Tani (1998) explores the possibility of establishing an
artificial version of Strawson's minimal self in a machine ... Tani,
however, in contrast to Strawson, makes it clear that the robotic self
he is designing is the result of physical interaction between the
robotic body and its environment...."

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

http://www.bdc.brain.riken.go.jp/~tani/
http://citeseer.nj.nec.com/tani98interpretation.html

[snip]

I become very leery eyed when I read this kind of stuff.

Bold unsubstantiated statements such as,

"Self consciousness emerges when the relation between
the subject (top down process) and the object (bottom
down process) becomes incoherent."

............
How best to integrate higher level processing with the lower
level processing does not require "mind" or "consciousness"
to be part of the equation, except as an emergent property
(a prediction of this kind of system) which would be a major
breakthrough in our understanding of what those terms mean
in a physical sense.

Not all that glitters is gold and my gut feeling is they have not
hit the gold yet.


Yes, of course, I agree with you about the excessive philosophizing
... silly word games again. However, what you might do is take a look
at the overall multi-leveled neural architecture of the robot, and its
method for integrating several top-down and bottom-up processes. Quite
interesting, I think. As I mentioned:

"... Tani's dynamical systems approach is the meat ..."

Recurrent NN's, self-learning, look-ahead prediction, visual
targeting, memory rehearsal, open-loop vs closed-loop predictive
sequencing, chaotic dynamics. Goes beyond just philosophical limit
cycling. The thing actually has wheels, too.

"dan michaels" <dan@oricomtech.com> wrote in message
news:4b4b6093.0402120832.42a777fc@posting.google.com...
"JGCasey" <kjcasey@hotkey.net.au> wrote in message
news:<402af50d_1@news.iprimus.com.au>...
"dan michaels" <dan@oricomtech.com> wrote in message
news:4b4b6093.0402111226.7d496014@posting.google.com...
The following might be of interest to people interested in building
embodied AIs.

I found a reference in the following cog.sci paper to Tani's robotic
work, mentioned below. The approach involves recurrent NN's and
self-learning schemes, integration of bottom-up perception and
top-down "subjective" prediction processes, in a real robot in a
real-world environment. The philosophical citation is just for
reference. Tani's dynamical systems approach is the meat. This
approach attempts to obviate the need for the homunculus.

http://www2.canisius.edu/~gallaghr/tics2000.html

"Philosophical Conceptions of the Self: Implications for Cognitive
Science", Shaun Gallagher, 2000.

"... Tani (1998) explores the possibility of establishing an
artificial version of Strawson's minimal self in a machine ... Tani,
however, in contrast to Strawson, makes it clear that the robotic self
he is designing is the result of physical interaction between the
robotic body and its environment...."

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

http://www.bdc.brain.riken.go.jp/~tani/
http://citeseer.nj.nec.com/tani98interpretation.html

[snip]

I become very leery eyed when I read this kind of stuff.

Bold unsubstantiated statements such as,

"Self consciousness emerges when the relation between
the subject (top down process) and the object (bottom
down process) becomes incoherent."

............
How best to integrate higher level processing with the lower
level processing does not require "mind" or "consciousness"
to be part of the equation, except as an emergent property
(a prediction of this kind of system) which would be a major
breakthrough in our understanding of what those terms mean
in a physical sense.

Not all that glitters is gold and my gut feeling is they have not
hit the gold yet.


Yes, of course, I agree with you about the excessive philosophizing
... silly word games again. However, what you might do is take a look
at the overall multi-leveled neural architecture of the robot, and its
method for integrating several top-down and bottom-up processes. Quite
interesting, I think. As I mentioned:

"... Tani's dynamical systems approach is the meat ..."

Recurrent NN's, self-learning, look-ahead prediction, visual
targeting, memory rehearsal, open-loop vs closed-loop predictive
sequencing, chaotic dynamics. Goes beyond just philosophical limit
cycling. The thing actually has wheels, too.

The practical model building was of interest. The idea that a robot
develops a "self" in the case of the Tani robot may have been inspired
by the fact that we become aware of otherwise unconscious actions
when something does not go as expected. However I doubt my
computer develops a "self" every time it gets an interrupt signal that
it has to "pay attention to" no matter how sophisticated or complex
that "attention" is. Computer people use "mind" terminology all the
time but they don't get confused by it.

Because I know people who have been diagnosed with schizophrenia
I was going to comment on the other reference you gave, "Philosophical
Conceptions of the Self:", but haven't had time to read in properly.

Researchers such as Jun Tani might make discoveries and come up
with models that may lead to a better understanding of the brain and
the "mind".

--
John Casey


Brain Science Institute, RIKEN

In article <402bf450_1@news.iprimus.com.au>, JGCasey
kjcasey@hotkey.net.au> writes

"dan michaels" <dan@oricomtech.com> wrote in message
news:4b4b6093.0402120832.42a777fc@posting.google.com...
"JGCasey" <kjcasey@hotkey.net.au> wrote in message
news:<402af50d_1@news.iprimus.com.au>...
"dan michaels" <dan@oricomtech.com> wrote in message
news:4b4b6093.0402111226.7d496014@posting.google.com...
The following might be of interest to people interested in building
embodied AIs.

I found a reference in the following cog.sci paper to Tani's
robotic
work, mentioned below. The approach involves recurrent NN's and
self-learning schemes, integration of bottom-up perception and
top-down "subjective" prediction processes, in a real robot in a
real-world environment. The philosophical citation is just for
reference. Tani's dynamical systems approach is the meat. This
approach attempts to obviate the need for the homunculus.

http://www2.canisius.edu/~gallaghr/tics2000.html

"Philosophical Conceptions of the Self: Implications for Cognitive
Science", Shaun Gallagher, 2000.

"... Tani (1998) explores the possibility of establishing an
artificial version of Strawson's minimal self in a machine ...
Tani,
however, in contrast to Strawson, makes it clear that the robotic
self
he is designing is the result of physical interaction between the
robotic body and its environment...."

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

http://www.bdc.brain.riken.go.jp/~tani/
http://citeseer.nj.nec.com/tani98interpretation.html

[snip]

I become very leery eyed when I read this kind of stuff.

Bold unsubstantiated statements such as,

"Self consciousness emerges when the relation between
the subject (top down process) and the object (bottom
down process) becomes incoherent."

............
How best to integrate higher level processing with the lower
level processing does not require "mind" or "consciousness"
to be part of the equation, except as an emergent property
(a prediction of this kind of system) which would be a major
breakthrough in our understanding of what those terms mean
in a physical sense.

Not all that glitters is gold and my gut feeling is they have not
hit the gold yet.


Yes, of course, I agree with you about the excessive philosophizing
... silly word games again. However, what you might do is take a look
at the overall multi-leveled neural architecture of the robot, and its
method for integrating several top-down and bottom-up processes. Quite
interesting, I think. As I mentioned:

"... Tani's dynamical systems approach is the meat ..."

Recurrent NN's, self-learning, look-ahead prediction, visual
targeting, memory rehearsal, open-loop vs closed-loop predictive
sequencing, chaotic dynamics. Goes beyond just philosophical limit
cycling. The thing actually has wheels, too.

The practical model building was of interest. The idea that a robot
develops a "self" in the case of the Tani robot may have been inspired
by the fact that we become aware of otherwise unconscious actions
when something does not go as expected. However I doubt my
computer develops a "self" every time it gets an interrupt signal that
it has to "pay attention to" no matter how sophisticated or complex
that "attention" is. Computer people use "mind" terminology all the
time but they don't get confused by it.


Some seem to John.

Because I know people who have been diagnosed with schizophrenia
I was going to comment on the other reference you gave, "Philosophical
Conceptions of the Self:", but haven't had time to read in properly.

Researchers such as Jun Tani might make discoveries and come up
with models that may lead to a better understanding of the brain and
the "mind".

--
John Casey


Brain Science Institute, RIKEN






--
David Longley

Not all that glitters is gold and my gut feeling is they have not
hit the gold yet.


Yes, of course, I agree with you about the excessive philosophizing
... silly word games again. However, what you might do is take a look
at the overall multi-leveled neural architecture of the robot, and its
method for integrating several top-down and bottom-up processes. Quite
interesting, I think. As I mentioned:

"... Tani's dynamical systems approach is the meat ..."

Recurrent NN's, self-learning, look-ahead prediction, visual
targeting, memory rehearsal, open-loop vs closed-loop predictive
sequencing, chaotic dynamics. Goes beyond just philosophical limit
cycling. The thing actually has wheels, too.

The practical model building was of interest. The idea that a robot
develops a "self" in the case of the Tani robot may have been inspired
by the fact that we become aware of otherwise unconscious actions
when something does not go as expected. However I doubt my
computer develops a "self" every time it gets an interrupt signal that
it has to "pay attention to" no matter how sophisticated or complex
that "attention" is. Computer people use "mind" terminology all the
time but they don't get confused by it.

As always, I'm interested in useful ideas for building AI's, and
couldn't care less if these things are conscious of self, have a mind,
or whatever, or what not ....

dan@oricomtech.com (dan michaels) wrote in message
As always, I'm interested in useful ideas for building AI's, and
couldn't care less if these things are conscious of self, have a mind,
or whatever, or what not ....


Excellent: comp.ai.philosophy vs sci.cognitive. My big problem in AI
development is what I dub 'multiple simultaneity'. To explain this
I'll describe a 'back burner' project running in my company.

dan@oricomtech.com (dan michaels) wrote in message
As always, I'm interested in useful ideas for building AI's, and
couldn't care less if these things are conscious of self, have a mind,
or whatever, or what not ....


Excellent: comp.ai.philosophy vs sci.cognitive.

The clerk seems to model the entire problem simultaneously, even where
it involves possible alternative processes. They seamlessly reclassify
the source data as a whole. This is why the machine works best if the
clerk sorts the documents into boxes and removes exceptional items.

The clerk seems to model the entire problem simultaneously, even where
it involves possible alternative processes. They seamlessly reclassify
the source data as a whole. This is why the machine works best if the
clerk sorts the documents into boxes and removes exceptional items.


And what was the learning curve for the clerk, and how can you get it
into the computer, once you figure out how to give the computer an IQ
equivalent of 100, or more? Your example illustrates well that MI
isn't as smart as a high-school or college graduate yet. [neither is
Tani's system, of course, but his system could probably navigate the
room your machine-clerk is booted up in].