Re: artificial intelligence

This is so damn long I don't even have the energy to proof it. It probably
has more typos than my normal posts. I wouldn't bother reading it if
someone else wrote it. But since I did take the time to write it and
finish it, I guess I'll post it anyway.... :) (I've thrown away many posts
in the past few days which I started to write when they got too long).

JGCASEY <jgkjcasey@xxxxxxxxxxxx> wrote:
On Dec 7, 1:52 pm, c...@xxxxxxxx (Curt Welch) wrote:
JGCASEY <jgkjca...@xxxxxxxxxxxx> wrote:
On Dec 6, 3:07 am, c...@xxxxxxxx (Curt Welch) wrote:
John:


When the source was located to the brain some may
still imagine it as a ghost in the machine but an
educated person in that area know that the mind
is correlated with neural activity


But a smart person will know that the mind is not
correlated with the neural actively, but in fact
IS the neural activity. This is the point you
never get, and the error you keep making that
makes me write all this, and the point I keep
trying to open your eyes to over and over without
getting anywhere.


If it's correlated with the activity, instead of
being an identity with the activity, then you are
making the argument there are two things here
instead of one. Where exactly is that other
thing and what is it made of?

Is light a particle or a wave? If you believe both
does that make you a dualist? A subjective experience
is what it is like for the nerve net, the visual
view of the net is what it is like for another
nerve net observing the physical make up of the
network in question.

That's mostly what I believe to be true.

You deny the two viewpoints

No I don't.

of the same thing
pretending you can "see" the brain from your
subjective viewpoint the same way you can see the
brain via your visual system.

Well, just to be clear, I was using the word "see" abstractly to mean
"sense". I wasn't trying to imply we were sensing a visual image of our
thoughts. But I think you understand that.

You fail to respond to the point I make about a
computer program not being able to see (without
a camera connected and pointing at them and the
intelligence to make the connection) the very
logic gates that is the hardware which makes
up the substrate of the program.

I think I missed that. I found it hard to read some of your replies
because you clearly were missing some major points at times and I didn't
have the energy to write another 1000 words to try and fix your mistakes.
That led me to just quickly skimming some of your posts.

Why do I keep saying you are a dualist? Because
you keep telling me there are two things here, the
neural activity, and the subjective experience -
two separate things when there are not two things
but one. See how that works - two things - dualism.

I think you choose not to understand. There is ONE
thing there from which there are TWO viewpoints.

Ok, let me talk to that some.

We can touch a dog, and we can see a dog. The sensory data comes into the
brain through totally different sensory paths. We have no problem
understanding the difference between touch and vision right? Not only does
the sensory data come into the brain by two different paths, but we are
very aware of this fact at the subjective level, right?

But at the same time, we never think of the dog as two different things do
we? We think of it as one dog? It just so happens we can both see the
dog, and feel the dog.

This is the same thing that happens with the brain. When we look at the
brain with our eyes (assuming someone was kind enough to open our our head
and give us a mirror so we could do it), we would know we were looking at
the brain with our eyes.

When we sense the brain with our thoughts, this is just another sensory
path that data about the brain is entering our brain. Our ability to sense
our brain behavior, is another sense, as real as our sense of touch, and
our sense of sight. Lets give this sense a name to make it a bit more
clear. I'll call it our subjective sensor.

It just so happens that we never call our ability to sense our brain
behavior a "sense" for reasons of historic misconception. So technically,
it's an invalid use of the English language to say I can sense my brain
with I'm having thoughts. But technically, it is a sense. Again, we can't
let the error made by the people who created our language mislead about
what is really happening.

The fact that we can sense our brain with many different sensors (ears,
eyes, fingers, subjective sensor), is what you are calling "two views".
That's an odd way to talk about "two senses" but it's ok by me if I can you
get to understand that is what is happening.

Now, just like when we sense a dog by different senses, when we sense the
brain by these different sensors, the same thing is happening. There is
only one brain there being sensed by two different sensory paths.

But here's the problem. It doesn't seem like "one brain" to us in the way
it seems to be "one dog". It feels far more disconnected that it. It
feels like two different things.

The brain fails to correctly solve the sensory modality integration problem
in this case. It works fine for dogs, and everything else in the world,
but it fails to solve it for the brain.

This brings us back to my theory that sensory integration is solved by the
brain on the basis of temporal correlation of sensory data events. The
brain will wire itself so that the feel of a dog, and the vision of a dog,
will map to the same internal representation because these things are
highly correlated. The different sensory signals keep happening close
together in time.

This never happens with the case of us sensing our brain because there
never is any correlated data for the brain to merge. We can't feel our
brain vibrate in a given spot every time we think of a dog for example. So
our subjective experience of "dog" never gets correlated to any sound, or
vision, sensory data. Because we can't see, or hear, our brain working,
the brain never forms a single internal representation for "our brain" like
it does for "the dog".

As such, our concepts of "brain" never get linked though the normal
automatic subconscious (if you wish) processing the brain normally does to
create an internal identity with "subjective experience".

That why we don't experience our understand of "brain" and "subjective
experience" as being the same thing, like we think of "vision of dog" and
"feel of dog" as being the same dog either way.

The reason people are dualist, (by the way I label everyone - not the real
definition of dualists) is because they have been able to understand that
the separation of subjective experience from physical stuff like the brain
is just an illusion created by the brain's lack of access to correlated
sensory data.

As I've said many times, I'm quite sure that if you gave the brain access
to correlated sensory data, it would form the connection for us using it's
normal system of self wiring to merge the subjective experience of some
thought, with the physical sensation of our brain operating as observed
with some form of brain scanning technique. With enough exposure to such
data, the subject would see the subjective experience as just another
physical sense of their brain hardware.

It is perfectly understandable if you want to think
it through in terms of a program running on a computer.

I think your understanding of computers is flawed. In your your other post
when you used this analogy, I wrote a long reply and then got tired of
trying to explain something to you I've explained to you many times in the
past and just deleted it.

Many people use this hardware/software analogy to talk about brain/mind.
But in fact, I'm quite sure it evolved the other way around. Our lack of
understanding of the brain, is what created the brain/mind abstraction and
then we copied that flawed abstraction when we create the hardware software
abstraction for computers.

By completing the circle, and trying to use the hardware/software
abstraction as a _justification_ for why brain/mind is _valid_ you have
just closed the loop on a circular definition.

For example, in another post, you said something like software can't exist
without hardware, but hardware can exist without software. This seems
valid on the surface, and it seems to show that software is not hardware -
they are different things. But it's a huge error in your thinking.

It's impossible to remove the software from a computer without removing
hardware. How do you remove the 0's and 1's in the memory of the computer
without removing the actual memory chips? You can't. No matter what you
do, the machine always has software in it.

In addition, to change the software, we do it by adding or removing
elections from the memory cells. We physically change the hardware to turn
the old machine design, into a new machine design.

There is no "software" there. There is only hardware. The concept of
software is again, a problem like Beauty in the Mona Lisa. It never was "in
the hardware". It's in the eye of the beholder. It's in the behavior of
the person looking at the computer and talking about the "software in the
machine". This is just as true about software, as it is about Beauty.

Why I stick a CD in my computer, I've added more hardware to it. When we
load the "Software" from the CD unto the disk, we have caused the disk
drive to be physically changed to match the patterns which are on the CD.
The "software" didn't move. The pattern was just duplicated.

If I give you a blueprint to build a house out of wood, and a pile of wood,
and you follow that blueprint and build the house, do you think of that as
"loading the software from the blueprint into the wood"? You should,
because it's the exact same type of physical process at work as when we
load software from a CD-ROM into a computer. We have simply taken a supply
of electrons from the power company and built a lot of different piles of
electrons according to the "blueprint" on the CD-ROM.

When you think of yourself as being "software" running on the "hardware"
you are making the exact same error as thinking "Beauty" is located in the
paint, or that "software" is located in the computer. It's located in the
behavior of the brain that is observing hardware, not in the hardware.

But do you grasp how tricky this all gets when we try to observe our own
hardware with our own brain?

If Beauty is not in the paint, but in the brain of the person looking at
the Mona Lisa then what happens when I look at you, instead of looking at
the Mona Lisa? My brain learns to recognize you, but the patterns of
behavior you create. When I see the right patterns of sensory data, my
brain identifies the pattern as "John". So we have a dualistic issue here.
We have you (which is the hardware - your body), and we have my brain's
representation of "you" (which is the sensory patterns my brain has
identified as being "John").

If we could clone your hardware and create a new body which acted in the
same way, I wouldn't be able to tell it wasn't you. The second body would
create the same sensory patterns as the first body, and my brain's pattern
recognition hardware would classify the sensory patterns as "John". I would
think it "was" John because of this.

So we have two very distinct concepts of what "John" is. We have the
hardware which is you (your body), and we have the behavior of any brain
which has been configured to recognize the sensory patterns which the brain
will identify as "John".

When you talk about "you" being the "software" running "on the brain" you
are not talking about your hardware at all. You are talking about your
hardware's ability to recognize the sensory data patterns created by your
hardware. But what is "your hardware's ability to recognize John"? It's
the behavior of your hardware. So, when you think of yourself as being
this software, what are you thinking about? You are thinking about how
your brain has the power to recognize "John sensory patterns". You are
talking about your physical brain's power to recognize it's own image in
the mirror (so to say).

I understand completely that these two "views" exist. I've talked about
them more times than I can count.

But what is the bottom line of all this? What does it reduce to? It
reduces to the fact that you are a physical brain, with the power to
recognize sensory patterns created by your physical brain.

Ill get t

How can a RL program get to learn it is a machine?

I can't tell why you are asking me this. I understand the answer
completely but I don't know if you are asking me to explain it or what?

And before it does that what is it experiencing, that
is, what does it really know directly? Certainly not
the hardware it runs on. There is nothing magical or
even difficult to understand here.

I think it's very difficult to understand and I'm not seeing any real
evidence you understand it to the level I'm trying to get you to understand
it.

If I touch a dog, do I not "know the dog"? Just because I've never seen
the dog (lets say because I'm blind) does that mean I do not know the dog?

Would you say I do not know the hardware of the dog because I've only
touched it instead of seeing it? I doubt you would. But you are saying
this when you try to talk about the brain and your subjective sensory power
- your brain's ability to sense itself internally. You continue to talk as
if sensing the brain is not sensing the brain when that sensor in use is
your subjective sensor.

How much does an RL program "know" about the hardware it's running on?
Very little. It certainly doesn't know it's made up of transistors. How
much do you "know" about a dog by touching it? Again, very little. You
certainly don't know it's made up of cells with DNA in them. But yet, you
seem to want to say that you "know" the dog, but you "don't know" the
brain.

That makes no sense to me. If we know aspects of the dog because we know
the patterns it detects, we also know aspects of the brain, because of
patterns it detects.

If you want to grasp what it's like to drop all
these old soul-based dualistic beliefs, you have to
learn to see the mind as an identity with the neural
activity, and not something correlated with the
neural activity.

There are two observations that correlate and yes the
evidence is they are the SAME THING. That you are no
more than the behavior of a vast assembly of nerve
cells. But you want to insist that there is only one
type of behavior. That the dancer only does one dance.

I think mostly this idea is consistent with what I believe.

But my issue, is that you seem to want to believe that "you" are a pattern
of behavior, and not hardware. You think that "you" are the

However the evidence is that conscious behavior comes
and goes. Just like that waving hand you like to talk
about, the amount of waving can vary from violent to
none at all. It probably involves some kind of short
term memory and is closely associated with attention.

You do nothing but create huge amounts of confusion by talking like that.

You keep talking about the patterns you (as an observer) are able to
recognize in the hardware, instead of talking about the hardware.

Hardware doesn't come and go so easily. It has quite a bit of persistence
in the universe.

The foundation of everything is the _hardware_. The hardware _always_ has
behavior. It can't stop having behavior. Whether the neurons are firing,
or they are all quite, the hardware is still there and still producing a
constant and continuous flow of behavior.

If you choose to define "consciousness" as one class of behavior (which is
all you have done above), then of course it comes and goes. At some point,
the hand is waving, at the next it's not waving - but the hand is always
there, and nothing change about the hand other than the patterns you, the
observer, noticed in it.

Likewise, you are using the word "consciousness" to mean nothing more than
"your perception of a pattern of behavior in the conscious device".

So, if we were to debate what consciousness is, do you understand what this
means your answer would have to be? It's nothing more than some arbitrary
pattern your brain has learned to recognize in the sensory data. It's some
arbitrary pattern recognition circuits wired in your brain, which defines
what consciousness is for you.

So what does it mean when I say my networks are conscious? Well, if your
"consciousness recognizing patterns matcher" doesn't see the right pattern,
you will say it's not conscious. Which is all fine an dandy. But stupid
to spend endless hours debating and stupid that you dont realize this and
just say: "Consciousness is what I want it to be, and I choose to say it's
not in your networks" - end of story.

But yet, you never seem to talk as if this is what consciousness is (a word
you defined for the fun of it). You seem to want to talk as if it's a well
known universal pattern.

The subjective view changes as the neural activity
changes for the subjective view is determined by the
current behavior of the neural net.

The "subjective view" IS the current behavior of the neural net. It is not
"determined by" the current behavior of the neural network. I don't know
why you want to keep writing all these things that imply it's not an
identity.

This would be true for any program running on a computer.

The world view of a computer program depends on how
it works not on the logic gates it runs on.

Well, yes, we can make different hardware "act" the same way. I of course
agree with that - especially in the limited domain of digital computers
where we define "act the same" as producing the same list of discrete state
changes.

But to try and say the behavior of the computer doesn't depending the logic
gates it runs on it just totally wrong. OF course it depends 100% on the
logic gates it "runs on".

Again, this gets back to all I wrote above about how I think you are too
vested in the idea that software is not hardware. At best, as I said
above, the hardware is the hardware, and the software, like Beauty, is not
in the hardware at all, but in the brain of the person observing the
hardware.

You seem to want to keep talking about the wrong hardware in the picture
but not understand you are doing it. You are talking about what the
hardware of the observer is doing instead of talking about the thing
observed.

There are two steps away from believing in soul-based
dualism and you have only taken one of them. The first
is to believe subjective experience is caused by
neural activity. The second step, is to realize
neural activity IS subjective experience.

Right. Just as the contents of a computer program makes
up the content of the subjective experience (viewpoint)
of the program not the physical hardware it runs on.

That's just wrong on so many levels. By saying that, you are implying that
when you watch software run on a computer, the "viewpoint" of the machine
is in your brain, instead of in the computer hardware. It makes no sense
to say that. Either talk about the computer, or talk about your brain
observing the computer and stop confusing the two.

The stuff you call "software" is not "in the computer" - it's in _your_
physical brain's interpretation of what the computer is doing.

The computer is the physical hardware, and nothing else. It does what it's
doing because it's restricted by the laws of nature to do that.

The fact that you choose to talk about it as "software running on the
computer" is a behavior of your brain, not a behavior of the computer. The
fact that when you look at the computer, you see "MS Windows Software" is
not a behavior of the computer. It's a behavior of your brain recognizing
"MS Windows Software" sensory patterns.

What I feel you are not grasping, is that the hardware is the lower level
foundation of all this, and if you want to talk about the "software running
on the computer", you have switched the discussion from the hardware, to
the higher level abstraction of "software". But when we translate that
higher level abstraction back to the hardware, we find the true subject of
your words is no longer the computer at all, but that it's your own brain
you are talking about instead of the computer. But yet you keep thinking
you are talking about the computer because that's what we pretend we are
doing when we talk about software. Just like we pretend that Beauty is in
the painting when it's not, it's in the eye of the beholder - or more
accurately, the brain which is beholding the paint.

It is the neural ACTIVITY not the neurons that make
up subjective experiences. Without the activity there
is no observing taking place of anything. Observing
is a behaving thing not a static thing.

Well, see, you don't seem to be grasping that when you switch to talking
about behavior, you have switched the true subject of your words from the
observed, to the observer. When you start to talk about the behavior of
the neurons instead of the neurons, you are in fact talking about the
behavior of your own brain while it observers, or thinks about, the
behavior of other neurons someone in the universe.

When you turn this confusion in on itself, and you talk about you observing
yourself (your own "consciousness"), it creates the illusion that the
"self" you are talking about is somehow separate form the hardware - the
illusion that there is some justified separation of hardware from behavior.
It gives you the false liberty to talk about "you" being something other
than the hardware because "you" are the behavior, instead of the hardware.
But all that happens only as a trick of language and reference.

When you look at yourself, and think about what you are, there isn't
something special there happening called "consciousness". There is just
hardware seeing itself in a mirror which is no more odd, or special, than a
digital camera taking a picture of itself using a mirror.

It's just physical signal processing hardware which has created a rather
direct feedback loop.

You got the first one, but you can't for the life
of you seem to grasp the second.

No you can't grasp that there are two viewpoints of
the same thing.

I have no problems grasping all this complexity.

What I have problems with is grasping how much of it you actually
understand.

It doesn't feel like you understand it all by how you keep trying to use
this hardware/software analogy.

It could be you grasp it all just fine and the only problem here is my
ability to grasp what you understand because you write in ways that seem
inconsistent to me.

It is physically impossible for a
program to directly see the hardware it is running on.
It must work that out via its input the way we did.

Ok, you clearly don't understand my points because you haven't though it
through enough yet. You are still working from too high a level and just
ignoring the details.

What does "to see" really mean? Lets use the digital camera as an example
to remove all the complexities of what the brain does. You might accept
this or reject it by saying the camera is not conscious so it's not a valid
example. But I'll run with it anyway.

I would say that when the camera takes a picture of a dog, and the data is
transferred to the memory, it has "seen" the the dog.

But there was a long internal chain of causality there as well. The light
came from the sun, and hit the dog, the dog generated some light which went
into the license, and hit the sensor, which caused a different electrical
levels to be transferred over the wires from the sensor to the processor,
and then into the memory.

Why, in this long chain of events, was the "dog" the only thing "seen" by
the camera? It's not. It's just how we choose to talk about it.

In fact, everything in the chain was "seen". The camera saw the sunlight,
the dog, the line of the camera, the sensor, the wires. It "saw" all those
things. It did not just "see" the dog.

If I look at fish swimming in a fish tank, do I not see the fish, the
water, the glass, and the air between me and the fish all at the same time
all mixed together in the same sensory signal? But why does it stop at the
air? Does not the higher levels of the brain also not "sense" the eye, and
the rods and the cones, and the optic nerve because all these things are as
much in the causality path as the fish and the water are in the causality
path?

We do not think of it as "seeing our optic nerve" when we look at fish, but
if you think about it, you realize this is actually what is happening.

Now, you can attempt to ignore this fact by taking a behaviorists sort of
stance and saying that "seeing" is what starts at the end, and is the
combined behavior of the entire "eye/nerve/brain": system and you could
argue that it makes no sense to say the "brain" is "seeing" the optic
nerve.

But think about this. If we have a hierarchy of detectors in the visual
cortex with one level detecting center surround patterns in the data form
the eye, and a next level detecting some sort of line orientation in the
center surround detectors, aren't the line orientation directors in fact
"seeing" the center surround detectors and not "seeing" fish?

Isn't it much higher in the processing that some part of the brain detects
patterns in the behavior of lower level parts of the brain and "sees" for
the first time the "fish" pattern? And when that part of the hardware is
"seeing" fish, is it not just as true that it's "seeing" the activity of
the other neurons as much as it is "seeing" fish?

When we see fish, are we not also "seeing the light produced by the fish"
instead of "seeing" the fish? Isn't that why when we look at a film, or a
photograph, we seem to be "seeing" fish even though there are no fish
there? It's because we are actually "seeing" light patterns.

But it doesn't stop there, it goes all the way into the brain up to the
point where the specific hardware in the brain which is assigned the task
of "seeing the fish pattern" is located. That hardware, is in fact not
seeing fish, or light, it's seeing the behavior of other neurons in the
brain - it's seeing the brain.

When we see fish, we are in fact, seeing our own brain, not fish.

If we had the power to activate those lower level sections of the brain to
make them again form "fish patterns" we would suddenly have a subjective
experience we would describe as "seeing fish". But in this case, it was
clear our report of "I see fish" was not correct. The correct report should
have been: "I see fish patterns in my brain".

Do you see how all this works? The higher level detectors in the brain are
seeing the lower level detectors at the same time we are seeing things like
fish, and dogs and everything else.

Does it "feel like" we are seeing the brain instead of seeing fish? Nope,
because we have no understanding of what it would be like to see fish
directly - all we have ever "seen" in our life is our own brain. We just
talk about that signal processing and pattern matching behavior of our
brain as "the subjective experience of seeing fish".

Until you grasp the idea there is only one thing
here instead of two, I will still call you a dualist.

And until you overcome your need for others to put down
as fools you will continue to fail to see that their
view on that subject may be exactly the same as yours
except they go beyond the "just is" to "the how can
we do it" and the "nothing but" to the "in what way
is it nothing but".

The how can we do it is trivial. Any signal processing system which has
the power to detect sensory patterns is already doing the same fundamental
thing, and the only thing, the brain is doing. AND gates in logic circuits
are already having what you want to call "subjective experience" because
they have the power to detect sensory patterns.

The fact that you, and a bunch of other people in this group can't grasp
that is preventing me from talking about the real problems that need to be
talked about and solved. I'm dealing with 1st graders in terms of what you
seem to be able to understand when I want to be working on the college
level problem.

The brain is nothing more than a large network of temporal sensory pattern
detectors which translates sensory data, into effector actions (arm and leg
and jaw motions). The major data flow path is sensory data to effector,
but it makes use of tons of minor feedback paths as well.

The other very important aspect is that the reaction network is not static.
It's constantantly changing it's logic - it's constantly changing which
sensory patterns it will react to. The major component of how it changes
is what I keep talking about as reinforcement learning. Learning is
nothing more than changing the logic of the temporal reaction network over
time.

The important things to solve, are how to build a large parallel temporal
reaction network, and how to make it change over time under the direction
of reinforcement learning. This is the problem to solve and the one I
desperately would like to find some people to talk about it with and work
on it together. But instead, I get to talk to the 1st graders and try to
get them into the second grade.

There is no mind body problem. There is no question about what
"consciousness" is. There is no question that when we sense private
thoughts, we are sensing our own brain. The "content" of our subjective
experience is the set of femoral sensory patterns our brain is able to
detect. Every human brain is wired a little differently, which means every
brain is able to detect a different set of sensory patterns, and as such,
every human brain has a different "subjective experience". But any network
of detectors is able to have some "subjective experience" just because it
is a signal network of doctors. So if you can call a human brain conscious
because it has the ability to detect sensory patterns, you should call just
about everything "conscious" at some level because all matter acts as
sensory pattern detectors to some minor extent. But this idea is a high
school concept and I can't tell that I've got any of you out of the first
grade yet on correctly conceptualizing the scope of what we dealing with
here and what we are trying to do as we try to build a machine which
duplicates human behavior.

I'd love to talk more details - but I can't grasp that anyone here
understands the details I need to talk about - so I try to get you up to
speed instead. And so far, I've failed to make much progress at all.

--
Curt Welch http://CurtWelch.Com/
curt@xxxxxxxx http://NewsReader.Com/
.

Relevant Pages

  • Re: What part of the brain is conscious?
    ... Ok, if the effect you keep calling consciousness is just neural activity, ... consciousness when the answer never will be found in the hardware). ... of his brain because he's as confused about this as you are. ...
    (comp.ai.philosophy)
  • Re: the explanatory gap
    ... The brain is part of the territory. ... that can be observed, is physicality action, and the only thing observation ... neurons, ... You don't grasp how important correlation is to our perception system. ...
    (comp.ai.philosophy)
  • Re: Does Searles "Chinese Room" argument imply that consciousness is non-scientific?
    ... Einstein of the brain will work it out and you will be able ... It doesn't seem to be neural activity ... It's blueness, ... When we play a video game we find ourselves getting sucked into the reality ...
    (comp.ai.philosophy)
  • Re: type writers and tooth aches II
    ... > is because there is a real "sense of gap" as you call it. ... environment via a sensory system that feeds the brain. ... > Temporal correlation between ...
    (comp.ai.philosophy)
  • Re: reinforcement learning and conscious pleasure
    ... response becoming more likely is called a reinforcer or reward ... if it's not causing learning, ... Not sure what you mean by "types of hardware". ... often like to think of the brain. ...
    (comp.ai.philosophy)