Re: Does Searle's "Chinese Room" argument imply that consciousness is non-scientific?

JGCASEY <jgkjcasey@xxxxxxxxxxxx> wrote:
On Oct 2, 4:46 pm, c...@xxxxxxxx (Curt Welch) wrote:

Of course conscious behaviors are special. Without
them we would not be human and could not direct our
learning processes. Think how you are conscious of
learning to ride a bike which later becomes a reflex
behavior without conscious behaviors being required
except to select the reflex process.

This seems to be a good example of how you are thinking that shows a lot
about how you view all this.

There is nearly an infinite number of ways to look at what might be
happening in the head. Our job when trying to solve AI is to find the one
abstraction which is as simple as possible, but still explains all the
objective data we have available.

The concept of our conscious self controlling our body so that the
subconscious can learn is a simple and obvious example of how we use the
concept of consciousness in our every day language. But our every day
language has evolved as a group creation of society over 1000's of years.
It was created by people who thought we had a soul and who had no
understand of modern science. It was created by idiots without access to
any of the information we have today. It's one of the worse abstractions
you could possible pick to try and explain what the brain is doing, or to
understand how it works.

But yet, you wrote the above, as if talking about our conscious awareness
and control was an obvious part of the brain. It's a fabrication - and a
bad one at that.

Of course, if we build human level AI, the robot has to be able to learn to
ride a bike. And it has to happen like it does when a human learns to ride
a bike. But what is actually happening here? Don't we have better words
to talk about it all? Aren't there some abstractions that come closer to
the truth about what happens in an event like that?

Why for example, might someone get on a bike and start trying to learn to
ride it? It's like that by that point in their life, they already have a
large set of language behaviors. Both the ability to recognize ideas
expressed in language and respond to them, and the ability to generate
language ideas.

Might someone be getting on the bike in the first place as a reaction to
language? Such as, a friend or parent might have suggested to them that
they should try to learn to ride a bike? Or maybe, they say someone else
riding a bike and produce words of their own such as "I should learn to
ride a bike"?

It's likely that they are simply reacting to fairly high level patterns in
their environment - such as words, either spoken by someone else, or spoken
by themselves. And it's likely that are producing these behaviors in
reaction to those elements of environment, because they have been
conditioned to react in those ways by a long history of past experience
reacting to words.

And, as they try to ride the bike, is it not quite possible that more word
behavior, or high level concept behaviors close to words, will be used as
he first attempts to get on the bike and ride it? For example, he might
say to himself, or have a high level concept similar to words, which
represents the idea that they need to get on top of the bike, and put their
*** on the seat, and hold onto the handle bars. All this is at first, is
likely to be high level language like behaviors being generated in the
brain, but not being expressed externally with words. But the effect is
just the same as if he had spoken words, and then heard his own words, and
responds to those words in the same sort of ways he has learned to respond
to those words when someone else says them.

So, when we are first learning to ride, our actions are being generated as
responses to high level concepts similar to words which are being generated
by some part of the brain which is responsible for recognition, and
producing, fairly high level concepts and high level behaviors. The same
part that can recognise a command from a parent to sit down, and respond
correctly by sitting down. Only now, we are commanding ourselves to do
these things, like "get on the bike", "put your foot on the petal", "push
off to start moving". Just as if we were watching someone else try to ride
and we were trying to tell them what to do. But it's happening much faster
since we are talking to ourselves using internal neural pathways - but it's
still a high level language like level of both producing the thoughts and
responding to them.

But as we learn to ride, the behaviors become conditioned as reaction to
whatever sensory conditions are existing in the environment. So instead
the high level command and response pathways get augmented by lower level
stimulus response pathways.

I suspect it's highly unlikely that this happened at only two levels
(conscious and subconscious). I strongly suspect it's a large continuum of
interconnections at different levels. We just call the shorter sensory
response paths subconscious, and the longer, higher level, stimulus
response paths, our "conscious" reactions.

The simple fact that one of the behaviors we can produce is language
behaviors, and that one of things we can recognize and respond to in the
environment is language, means that we have this whole set of behaviors
which gives us a special dimension to our behavior. It's one thing for me
to look at a box, and not generate any language thoughts about - but to
just look at it. But what happens when I look at it and start saying
things in my head such as, "that's a black book; I wonder why it's there;
maybe my wife put it there; maybe I should ask here".

Those words are just a string of behaviors produced by the body like any
other string of behaviors, such as getting on a bike and riding down the
street. But yet, following social convention, we might say I was conscious
of the book, but I was unconscious of my bike riding? Why would we call
one string of behaviors conscious behaviors and another unconscious? It is
because we have no memory of what the feet did, but we have a memory of
those words we produced about the book?

I suspect it's more to do with the fact that when we produce words as a
reaction to some stimulus event, we see that as an act of being conscious
of the events that the words were about. I suspect also, that if we
produce a unique combination of words, it's easier to remember the words -
or it helps us later remember the event because there was more the event
happening in our head - there is both the memory of the stimulus event, and
the memory of the words that we spoke to ourselves about the stimulus
event.

I also suspect that when we produce these high level concepts that are
words, or very similar in level to words, and then we respond to those
words, that we call that type of internal behavior "conscious behavior".
So if my brain produces the sequence of high level concepts (aka thoughts)
relating to hunger, and a eating a sandwich, and then the thoughts of
making such a sandwich in the kitchen, and then we get up and go to the
kitchen and make the sandwich, we set that as a conscious reaction.

But, if my high level concept part of the brain was running though some AI
concepts, and I then got up and started walking, and ended up in the
kitchen standing in front of the fridge with the door open, but all the
time, the part of the brain which recognize and generates these high level
concepts, was busy with thoughts of AI, we say the high level behavior is
what we were conscious of, but the walking to the kitchen we we not
conscious of - simply because we didn't produce any words that triggered
our walk to the kitchen, or we didn't produce any words about the kitchen
or food, or why we were in the kitchen.

But, all this sort of behavior can be seen and explained with some very
simple types of brain hardware. First, the brain must have sensory
processing circuits that can recognize patterns of sensory data. It must
recognize that some patterns of sensory data is a dog, and others is a cat,
and other patterns it the words "look at that cat", and other patterns of
words, is the concept that someone wants me to look at a can in the
environment". It all starts with a large pattern matching network which is
able to uniquely recognize billions of different patterns.

Second, the brain must have the power to produce different sequences of
output behaviors. It must somehow, have the power to produce output
patterns.

And third, there must be something at work, selecting what output to
produce, based on some type of logic which triggers different output
patterns at different times (along with complex merging and prioritizing
and parallelism of different patterns happening at once, etc.

But once you have these sorts of basic brain powers, you can explain all
the things that are happening when we learn to ride a bike under what you
talked about as conscious control.

All the high level word-like concepts that our popping into our head about
the things we should be doing. But these high level concepts are nothing
more than internal brain behaviors which are being selected based on
whatever complex logic controls all our behaviors. It's just selecting
these private behaviors instead of selecting the external behaviors - like
arm motion.

But the brain is constantly parsing the flow of sensory data and has a
constant level of awareness of what's happening around it as it decodes all
this sensory data into the things that it represents in our environment.
And that whole set of things that our around us, is all input to the logic
which is selecting what behaviors to produce next.

But when it's producing these internal brain behaviors, those brain
behaviors are as much a part of the environment as what we are seeing with
our eyes. It's all part of the large context the brain is aware of - and
it's that large context that drives what behavior the brain will produce
next. So when we have thoughts like "I want to move my hand", and then the
hand moves, it's just because part of the logic of behavior selection is
wired to make the hand move after it senses that thought pattern show up in
the brain.

When we are thinking about something, by producing a series of words in our
head, we generally have no clue why those words just popped into our head.
They seem to be coming into existence with no cause. But we know that's
not true. Those words, or those thoughts, which are showing up in our
mind, are just behaviors produced by the brain, and the brain is using some
mechanical process to select those behaviors instead of selecting all the
other behaviors it could have selected.

But, when we have the thought "reach out and touch that", and then we reach
out and touch that, we see the "cause". We see that the thoughts happened
first in our head, and then the body responded. Because of that, we feel
like the thoughts were in "control" of the arm motion. But since we don't
know what was in control of the thoughts, we think of them as not being
under the control of anything. We thing they happen by our "free will".
Free will is just a fancy way of saying we don't know what controls our
behavior - and since it looks like nothing is controlling it, we think it's
free.

But in fact, for a materialist, we know that atoms just don't move on their
own. They move because some other atom pushed them. There is always a
causality chain to trace back because energy is conserved - it came from
somewhere and it went somewhere creating a causality chain.

So, you said it's human that we have the ability to consciously control our
learning, such as bike riding.

But what you labeled in this one simple example as "conscious behavior" I
see as nothing more than the brain producing language level behaviors, and
the brain responding to those same internal behaviors because it has
feedback loops that allow it to produce internal behaviors, and then react
to them as if those internal brain behaviors were part of the environment.
And I believe all our responses are there, as a result of being shaped by
operant conditioning.

I don't believe that the part of the brain which is recognizing, and
producing, these high level thoughts, is doing anything very different from
the part of the brain that's producing leg motions. The brain simply have
complex logic which causes it to produce these different behaviors in
response to different environmental contexts (where the internal state of
the brain is the environmental context it is reacting to - some of that
state is a direct representation of the state of the external world (what
the external sensors are picking up) but some is not - it's activated as
the result of feedback paths.

I think what you called conscious behavior or conscious awareness in this
one example, is just a label for our higher level ability of the brain to
use it's pattern recognition hardware to recognize language (syntax) and
language concepts (higher level language patterns), and for the brain to
use it's behavior selection system, to produce language behaviors. And I
think all this behavior, is a result of operant and classical conditioning.
That is, the hardware logic that determines which behaviors are produced
next, has all be tuned, and configured, through a long process of
conditioning.

There's nothing really special happening here, other than pattern
recognition circuits triggering behavior sequence generators. You can call
it important conscious behavior which makes us human, but I just see it as
a reinforcement trained reaction machine which includes the ability to
recognize and generate some fairly high level language like concepts.

In order to get human language like behavior, we might have to create a
special network topology so that part of this reaction system will be able
to form the section that recognizes, and triggers language behaviors which
is somewhat isolated from other sections which are more directly flooded
with with sensory data are more directly associated with the production of
behavior - but still, I see it as the same underlying technology making it
all work - temporal pattern recognition networks which are evolved/trained
by reinforcement learning and which also self adapt to the sensory patterns
which exist in the data.

I think using the aberrations of consciousness and unconscious behaviors
which is an idea dreamed up as a group effort by a society who believed in
Gods and souls, is the worst description of the brain we have. It's based
on the concept that if we don't understand why the brain produced behavior
(aka we don't see a linked temporal cause), it must have been produce by
the soul. We can do much better than that now. Behaviorism has looked
closely at raw object data, and found there was no soul. Or, you could
say, they found that the soul was just a mechanical process of behavior
selection based on stimulus events in the environment (where the things
that happen in our mind is as much a part of that environment as the things
that happens outside our body).

This large recognition and reaction network might well be built as modules
with some fixed pre-wired cross connects (aka topology) so a fixed about of
network ends up being applied to different parts of the behavior problem.
But I'm sure it's all one common technology making it work. And before we
can figure out what type of overall topology might be needed to closely
match the powers the human brain has, we first have to figure out how to
build one of these modules - something that can parse sensory data into
unique objects, something that can produce output sequences, and something
that can link the two, such that there is some sort of complex logic
mapping the current context as defined by the recognition hardware, to the
correct behaviors to produce for the current context. If we can figure out
how to build that type of hardware, and build it on a way that allows us to
expand it from small context/behavior problems to very large
context/behavior problem spaces, I think we will have basically solved the
problem of AI. And if you want to call the the part of the network
recognizing, and producing, language level concepts, the "seat of
consciousness", you can do so, but I think it's an odd way to talk and one
which we only use so much in our day to day lives because of long held
social traditions stated by people who had no clue what sort of magic
allowed them produce the behaviors they produce.

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

Loading