Symbols

Subject changed from "Is technology part of evolution?" and group set to
only c.a.p since I don't think the other groups care about this.

Traveler <traveler@xxxxxxxxxxxxxxxx> wrote:
On 07 Aug 2007 07:25:57 GMT, curt@xxxxxxxx (Curt Welch) wrote:

All I am going to add to what I've already written in this thread on
this subject is that a network cannot possibly be processing a symbol
if it does not "know" that it is processing a symbol. You are
projecting your private view of the process onto the machine. The
machine could not care less that you believe that a signal is a symbol
that represents an external phenomenon.


OK, not only do we seem to disagree on this, I can't even grasp what you
are thinking. So lets work on this.

For starters, what exactly do you think a symbol is?

Here's a simple definition from dictionary.com:

1. Something used for or regarded as representing something else; a
material object representing something, often something immaterial; emblem,
token, or sign.

If I write the word "AI" in the sand, I think we all agree it's a symbol.
Right? But when is a scratch in the sand not a symbol for you? What makes
one scratch in the sand a symbol, and another scratch not a symbol?

Is it only a symbol if there is a symbol processing machine responding to
it?

To me, the letter A printed on the key on my keyboard is a symbol. It's a
symbol because when that symbol is transmitted from the keyboard, to by
brain, through my eye, I will respond to it in a specific way. It's how I
respond to these things that define their meaning to me. I can see the
symbol AI in the sand, or on the screen of my computer, or written in pen
on a *** of paper, and all these marks are symbols that cause me to
respond in a similar way. They are the "same symbol" to me, because of the
fact that my body responds to all those symbols in a similar way. It's my
physical response that defines the meaning of those symbols to me. The
meaning of a symbols, is ALWAYS defined by how a physical system responds
to. This can make the meaning of the symbol AI very different to one
person than another. We could train a human to see the symbol AI as
meaning something odd like: "you are about to be bitten by a dog". To that
human, the symbol would have very different meaning. But it would still be
a symbol. It still represents something to that human.

If I see a footprint in the sand, it also means something to me. It means
there was a human walking on the sand. To me, the footprint represents
something. It represents the recent past presence of a human at that
location. The footprint to me, is a symbol. It's a symbol because it
represents something to me - because it causes me (aka my brain) to
physically respond in a special and specific way.

If you read books on information theory, you seem them use the word symbol
to talk about the encoding of information. It's encoded as a stream of
symbols being "received". Information theory doesn't limit the concept of
symbol to just humans receiving information. It makes no difference if
it's a computer or a human, it's still a sequence of symbols. Bits sent
over any communication channel are considered symbols according to any book
on information theory. By how it's used in information theory, signals are
streams of symbols until you move from discreet to continuous signal
formats. In other words, any type of discreet signal system, is made up of
symbols.

So what exactly is a symbols to you that makes you want to say your bit
processing programs are not symbol processing programs?

You did write the very odd phrase above:

... a network cannot possibly be processing a symbol
if it does not "know" that it is processing a symbol.

If this is a key part of your definition, then I guess I also have to get
you explain to me what you think "knowing" is. Is this going to get us
into the sticky area of consciousness which we never agree on? Is a signal
not a symbol unless it's received by a conscious agent and can computers
never be conscious agents to you? Is this issue of consciousness what
makes you attach some odd special significance to the word symbol that
makes it so hard for you to grasp what I'm talking about when I say spikes
are symbols?

"The operational closure of the nervous system." Look it up, again.

Well, all I find is your web page on it. :)

In that section you write for example:

"There is nothing in a neural spike that identifies its origin
from the point of view of a receiving neuron."

Of course there is. The nerve the spike travels on identifies it's origin.
To talk about the spike as if it were somehow separate from the path it
traveled on is absurd. You are mixing levels of abstraction very badly in
what you write to the point that you are creating total nonsense.

I agree completely with you that it's important for us to discover machine
designs that use sensory independent processing algorithms. That is, we
want to find one algorithm that works just as well for spikes coming from a
light sensor, as it does for spikes coming from a sound sensor. The spikes
are the same format no matter where they come from, and the processing
applied to the spikes should be the same no matter what signal we are
trying to process. The processing must be signal independent - which
simply means it's the same processing we use in the machine to process all
the sensory data signals.

What it DOESN'T mean, is that signals have no meaning! What it doesn't
mean, is that the signals don't each REPRESENT different, and unique,
things in the external. The spikes coming from the light sensor represent
light. The spikes coming from the sound sensor represent sound. To pretend
they don't is just crazy talk.

And if they are representations of something else (which both are in these
examples) then they are, by simple normal English definition from
dictionay.com, SYMBOLS.

So what is about this that you can't follow? How, in your way of thinking,
have you created a definition of the word symbol that isn't consistent with
everything I've written above?

And since you Dan, said you don't agree with my use of the word symbol, you
too might choose to weigh in here on how you define the word such that it
prevents you from understanding that all spikes in the brain or in any
spike processing network are symbols.

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

Quantcast