MORE DME/HOMUNCULUS-RKC

[From Bob Clark (930403.2200 EST)]

RE: Martin Taylor (930331 16:30)

Thank you very much for your prompt and thoughtful reply to my post of
930331.1420.

Before turning to your specific remarks, I'd like to state where "I'm
coming from." It seems to me that the present theory is incomplete in
certain respects, and would be much more useful if ways can be found to
improve it. Here I point out two places where it is incomplete:

* * * * * * * * * *
1. BCP, Chapter 5, Memory, pp 220f, and Figure 15.3 showing the two
position switches. Here the four possible combinations of the switches are
described in terms of "Modes."

After discussing these Modes, we find, BCP p 224:
"Note how skillfully I bypassed the question of *what* flips the memory
switch. "One" flips it! I plead guilty to obfuscation -- the model
obviously lacks some details which I am not now prepared to supply."

* * * * * * * * * *
2. In BCP, Chapter 13, Higher Levels, pp 173ff, Bill remarks:
"I must now account for choice of particular system concepts as ninth-order
reference levels, and I can't."

Also: "So I must say for the time being that this *is* my model of
behavioral organization, as far as it concerns the ongoing performance of a
competent adult human being. I must leave questions unanswered, hoping
that others will find this approach interesting enough to expand upon and
modify."

Then: "Another possible--even probable--source of ninth order reference
levels is *memory*."

Further: "The solution that I prefer for this problem involves a discussion
of learning of a particular type, and so will be presented later."

This "learning of a particular type" is, of course, the Reorganizing
System: genetically determined and operating outside the hierarchy with
Intrinsic Error providing the driving force. See BCP, Chapter 14,
Learning, p 182, and Glossary, p 285, Intrinsic Error.

[Incidentally, this system was originally proposed (as the "Negentropy
System") with essentially the same properties as the present "Reorganizing
System." It was proposed in order to account for observed changes in the
operation of the systems composing the hierarchy. It was regarded as
operating "outside" of the hierarchy -- without definition of "outside."]
* * * * * * * * * *

DISCUSSION of the above two items.
The "One" in the first item is regarded as existing, somehow, somewhere. I
am suggesting a more meaningful name, that will help identify the "items"
needed to accomplish the indicated results. Perhaps Decision Making Entity
(DME) is not the best name for this, but some equivalent seems to me
unavoidable. I have listed seven items [Bob Clark (930331.1420)] which
seem to me necessary for the DME to perform its switching function
effectively.

The DME is proposed in order to account for certain observable events
called, perhaps loosely, "decisions." Many of these are readily accounted
for in terms of the existing hierarchical structure, including pre-set
"choice-points," as discussed in BCP, Chapter 14, Learning, p 177f:

"Programs are fixed lists of instructions (reference levels for lower-order
systems in human beings) with choice point in the lists. Both memory and
present-time inputs are important elements. ... the same list of
operations remains in use, and event the subprograms may retain their same
organization. All that changes is the path followed through the network of
contingencies, ALL POSSIBLE PATHS BEING DETERMINED WHEN THE WRITING OF THE
PROGRAM IS FINISHED." [Capitalization by Clark.]

Although a mature adult may have adequate programs to cover all possible
situations, it seems unlikely. It seems especially unlikely for an infant,
where a major part of its learning consists in learning such programs.

Others may be accounted for in terms of the operation of the "Reorganizing
System," at least when an "Intrinsic Error" exists.

Operation of the Reorganizing System may account for those "decisions" when
an Intrinsic Error exists. "Decisions" made in the absence of Intrinsic
Error require other operations.

ยทยทยท

--------------------------
Turning to your post (930331 16:30):

Consciousness is a multidimensional experience.

I'm afraid I don't know what this means.

To me, "consciousness" refers to the condition of the perceptual systems.
If they are in working order, the individual is "conscious." I think this
is consistent with the following:

BCP, Chapter 14, p 200, "Consciousness consists of perception (presence of
neural currents in a perceptual pathway) *and* awareness (reception by the
reorganizing system of duplicates of those signals, which are all alike
wherever they come from.)"

Again from you:

"ME" is an element of consciousness.

Is "ME," then, a subgroup of perceptual signals assigned the label "ME?" In
the same sense as the "tree in the yard?" Is it always passive? Is it
sometimes active? In what manner, subject to what conditions. if any?

USER'S VIEWPOINT.

From you:
What I mean by a "User's" viewpoint is that you can take account of only
the signals accessible at that point.

"Point" in the hierarchy, or "Point" in time, or both?

The User's viewpoint of an ECS is not that of a person within whom the ECS
operates.

If "ECS" means "Environmental Control System," I don't understand the
preceding statement.

It is consideration of what is accessible at some point within the ECS,
often the perceptual signal, but possibly one of the other signals.

If "It" refers to the "User's viewpoint," I don't understand this statement
either.

Your DME does not have a User's viewpoint of the action hierarchy.

Since I don't understand your definition of User's viewpoint, I cannot
comment.

It has access to signals from all over the hierarchy.

Yes, this is what I said.

What is a "decision" WITHIN the control hierarchy? etc

Your discussion here seems to consist largely of a description of the
ordinary operation of the hierarchy using its existing Choice-Points at
whatever levels may be required. I am concerned with situations in which
Problem-solving programs are, perhaps, incomplete, or otherwise unable to
provide needed solutions. But, at the same time, no Intrinsic Error exists.

....If you are talking about a Powers type of hierarchy, you must be
talking about the program level or above, because below this level there
are no choice points.

So I am suggesting a situation where there is no suitable "program"
available, with or without pre-existing choice points. Perhaps this is
rare, although at early stages of development there may be rather few
effective "programs." Here is where a DME may produce action before any
Intrinsic Error develops.

How does one "anticipate" intrinsic error? One can't even perceive it
when it does occur, according to Powers.

Do you agree that it is possible to "anticipate" some possible future
events? Do you plan your posts before sending them? In giving a talk, do
you plan for possible questions or interruptions? Is a toothache an
Intrinsic Error? Do you remember having one? Or any other Intrinsic
Error? Do you go to the dentist to avoid a future toothache? Do you take
action to avoid repeating a situation involving an Intrinsic Error?

You refer to my "1. Current Perceptions," and ask:

On what basis is this selection made?

Initially, it could be an unusual, attention-getting event. It could be
completely arbitrary. Perhaps, curiosity. This is only one item in the
seven I listed. They must be considered as they work together.

Also:

What is the perception the DME is controlling by means of varying its

choice of neural signals?

It is controlling its perception of the over-all situation as it relates to
an unexpected event. This includes its examination of those memories that
seem to be related.

And:

in what control loop?

In whatever control loop has a problem, but lacks a pre-existing
Problem-Solving Program.

After referring to my 6., you offer:

So the hierarchy is the environment on which the DME operates, exactly as
does the Powers reorganizing system? Your seven characteristics certainly
seem to indicate this. But how does the DME itself operate? Is it
controlling anything? If so, what can it be controlling but its own
perceptions?

Very pertinent and important questions.

First, the DME has a strong resemblance to the "Powers reorganizing
system." It differs in that it only operates with respect to those
perceptions that have its attention, whether selected arbitrarily or in
response to some attention-getting event. This is in contrast to the Reorg
Sys, which is in contact with all intrinsic signals all the time.

Second, as long as the established hierarchy has no problems, the DME need
not be active, but is capable of arbitrary action, perhaps "curiosity."

Third, its primary source of material to use as inputs to the hierarchy is
the contents of the memory. There is little discussion, in BCP or
otherwise, of the contents of the memory. But, after all, without the
memory there really is no hierarchy beyond the genetically determined
neural pathways. Note, again, the Reorg Syst output seems to be arbitrary,
if not entirely random, making no use of contents of the memory.

Fourth, after either the Reorganizing System or the DME acts, the DME can
review (in imagination) those actions as they were recorded in the memory.
They can then be described in terms of a "Problem-solving Program,"
complete with choice points. Indeed, this additional Program becomes
available for future use. It seems to me that this is pretty much the way
these Problem-Solving Programs come into being.

Thus the DME needs no hierarchy of its own. When there is a problem, it
provides assistance to the hierarchy based on selection from the contents
of the memory.

In our early work, Bill and I were both greatly concerned with possible
circularity or other recursive properties of the developing theory.

Thus, BCP, Chapter 14, Learning, p 182, "Throughout the development of this
theory, I have remained constantly aware of the "little-man-in-the-head"
problem, and have tried to avoid giving the reorganizing system any
property that depends on the operation of the very hierarchy that is
constructed by the reorganizing system. Whatever process is involved in
reorganization, it must be of such a nature that it could operate before
the very first reorganization took place, before the organism could
perceive any thing more complex than intensities."

The DME requires at least the beginnings of a recording function and the
formation of memories. These are included, as noted above and in BCP
Chapter 15, Memory, within the present theoretical structure.

My problem with the DME is that it seems to do the same kind of job within
the main hierarchy that the main hierarchy does in the outer world.

If I thought so, I, too, would reject the DME concept. Instead, I see it
as playing a critical part in the development of the hierarchy.

I hope that I have shown you where and why I think the DME concept offers a
useful extension of the original theory.

Thanks again for your interest, questions, and ideas.

Regards,

Bob Clark