DME & HOMUNCULUS - RKC

[From Bob Clark (930331.1420 EST)]

RE: Martin Taylor (930328 20:30)

Your Post reads:

Count me among the non-acceptors. Your DME sounds very much like the old
homunculus who sits behind the sensors and effectors, manipulating. How
does he work? Does he have his own little hierarchy?

OBSERVER'S VIEWPOINT (Bob's)

From this viewpoint, this post is equivalent to two decisions: 1) "Count me

among the non-acceptors" is equivalent to your (Martin's) having decided
that the First Person Singular does not refer to a Decision Maker; and 2)
the posting of your (Martin's) decision to the Net is equivalent to a
second decision.

USER'S VIEWPOINT (Martin's) REQUESTED
Who, or where, is the "ME" included in your (Martin's) post and involved in
creating it? Please explain your (Martin's) alternative(s), with or
without using PCT. Remember, I (Bob) am assuming a situation where both 1)
the established "nodes" or "choice points" are, for any reason, unable to
provide a "decision" and, 2) "Intrinsic Error is neither present nor
anticipated.

"HOMUNCULUS" -- THEORIST'S VIEWPOINT (Bob's)
My dictionary (Webster's Encyclopedic Unabridged Dictionary, 1989) gives:
"homunculus, n. 1. a diminutive human; midget. 2. a fully-formed,
miniature human body believed, according to some medical theories of the
16th and 17th centuries, to be contained in the spermatozoon."

I don't think you intend the term, "homunculus," to be taken literally per
the definition above. More important, in my posts I have tried to restrict
the capabilities of the DME to those without which it could not perform its
defining function: making decisions. Can any of these be omitted? Should
any others be added?

"HOW DOES THE DME WORK?" -- THEORIST'S VIEWPOINT (Bob's)
For the original suggestion, see Bob Clark (12052). Also the post to which
you responded, Bob Clark (930327.1945 EST). Other posts have attempted to
further clarify the idea.

Here is another attempt to describe the essential characteristics of a
Decision Making Process -- a DME in operation.

It is assumed that:
A. no built-in automatic branch-point is available
B. no Intrinsic Error currently exists.
C. no Intrinsic Error is anticipated.

I have tried to limit this description to those items without which
Decisions cannot be made. Thus the proposed items are:

1. Current Perceptions. The DME selects the signals to which it directs
its attention. They are selected from among the incoming neural signals
available. These signals are available for use as feedback signals if
needed.

2. Current Objectives (reference levels, etc), if not already in
operation, are selected from Recordings of past decisions, events, etc.

3. Past Perceptions -- Recordings (Memories). The DME finds Recordings
both by named addresses and by similarities of content. They may result
from simple "recognition" ("reminders"), or (more or less) extensive
searches for relevant material.

4. The Recordings found are examined for relevance and possible
application ("feasibility") to the Current Perceptions (Perceived
Situation).

5. The Recordings are further examined, by imagination, for anticipated
future effects as they relate to current, relevant reference levels
("objectives," "purposes," etc).

6. The entire hierarchy is available to serve as the Output Function for
the DME. In ordinary situations, only limited, selected portions will be
needed.

7. On the basis of the above examinations, etc, the DME selects and
activates a Recording. The DME's selection can be arbitrary. The
Recording selected can consist of revised and/or combined Recordings.

If this is a "Homunculus," so be it.

For Perceptual Control Theory of Behavior to be complete, it seems
essential to me that "Decision Making Activities" are included somehow.

In addition, I think that these elements are consistent with most, if not
all, of the ideas either stated or implied by Powers in BCP.

Finally, to repeat; Please let me know your (Martin's) procedures for
making ordinary decisions, and what part(?) of you (Martin) does it.

Regards,

Bob Clark

[Martin Taylor 930331 16:30]
(Bob Clark 930331.1420)

USER'S VIEWPOINT (Martin's) REQUESTED
Who, or where, is the "ME" included in your (Martin's) post and involved in
creating it? Please explain your (Martin's) alternative(s), with or
without using PCT.

Two questions and an assumption. I recognize the existence of consciousness
in me, and extend you the courtesy of assuming it exists in you. I have
no explanation of it, other than the simple presumption that its content
must be based on signals in the hierarchy, and that it is not itself such
a signal. Consciousness is a multidimensional experience. "ME" is an
element of consciousness.

The assumption: that this is a User's viewpoint. What I mean by a "User's"
viewpoint is that you are can take account of only the signals accessible
at that point. The User's viewpoint of an ECS is not that of a person within
whom the ECS operates. It is consideration of what is accessible at some
point within the ECS, often the perceptual signal, but possibly one of
the other signals. Your DME does not have a User's viewpoint of the action
hierarchy. It has access to signals from all over the hierarchy.

Remember, I (Bob) am assuming a situation where both 1)
the established "nodes" or "choice points" are, for any reason, unable to
provide a "decision" and, 2) "Intrinsic Error is neither present nor
anticipated.

What is a "decision" WITHIN the control hierarchy? It must be something
that happens at the program level or above (assuming Bill Powers' set of
levels). Below the program level, there may well be multiple means to
achieve any particular perceptual signal value, but the variation of means
must be caused by differences in the reactivity of the world. The increase
of difficulty (I sometimes say "impedance") of one lower-level control
may mean that a higher-level perception is brought under a control by
an entirely different set of actions. This is not "decision" as I
understand it. It is a natural consequence of there being a non-linear
system with more (in this situation) degrees of freedom for output than
there are perceptual degrees of freedom being controlled at a high level.

Something nearer "decision" can occur within the hierarchy below the
program level, when possible actions are played through imagination
loops in various ECSs. I suspect this happens all the time, and is
not switched. The effectiveness or otherwise of this imaginary control
may affect the real gain of different ECSs, resulting in different real
patterns of action when the imaginary control is actualized. Again,
there is the appearance of decision without any actual decision.

At the program level, "decision" is intrinsic to the level. It is the
nature of the program level to select among sequence reference levels.
There, decisions have an explicit place within the hierarchy, and I
would think that they would be accessible also at higher levels. But
that's pretty high in the hierarchy.

Now back to your quote: 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. The PIFs do not permit them.

How does one "anticipate" intrinsic error? One can't even perceive it when
it does occur, according to Powers. I don't think it is relevant to the
issue of the DME.

[HOMUNCULUS definition omitted]
I don't think you intend the term, "homunculus," to be taken literally per
the definition above.

No, of course not. One of the reasons that behaviourist psychology
became popular in the early years of this century was that people saw
that most of the 19th century psychological theories were recursive. To
explain what a human did, they in effect passed the results of sensory
processing to a "little man in the head" who decided which levers to pull
and push to make the muscles work. All the issues of the psychology of
the human were incorporated within the LMITH, and he was usually called
"the homunculus."

1. Current Perceptions. The DME selects the signals to which it directs
its attention. They are selected from among the incoming neural signals
available. These signals are available for use as feedback signals if
needed.

On what basis is this selection made? What is the perception that the
DME is controlling by means of varying its choice of neural signals?
"...for use as feedback signals..." in what control loop?

6. The entire hierarchy is available to serve as the Output Function for
the DME. In ordinary situations, only limited, selected portions will be
needed.

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? And if it is controlling its own perceptions, do not
the same considerations apply to it as to the main hierarchy: it is
a hierarchy of perceptual control systems, needing a sub-DME to make
decisions on its behalf, such as what signals in the hierarchy to
attend to?

If this is a "Homunculus," so be it.

Well, it still sounds like one, in that it solves an acknowledged problem
within the control hierarchy by replicating the problem at a new level.
The recursion, as with the original psychological conception of the
homunculus, is potentially infinite.

For Perceptual Control Theory of Behavior to be complete, it seems
essential to me that "Decision Making Activities" are included somehow.

Yes, but why must they be outside the control hierarchy? Isn't the
program level adequate? Remember that in the Powers system, perceptual
input functions may accept any neural signal as input, though in our
diagrams and analyses we usually consider only the perceptual signals
of the next lower level of control.

Finally, to repeat; Please let me know your (Martin's) procedures for
making ordinary decisions, and what part(?) of you (Martin) does it.

If I knew that, I would join the ranks of those making pronouncements
about the truth of the world, and I might be rich, into the bargain.

Look, my problem with the DME as an entity isn't a matter of faith that
everything can be solved within the main hierarchy (though I like to
think that true, and is one reason I continue to think of local
reorganization instead of postulating a separate reorganizing system).
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. That means that the DME must need its own DME, which needs
its own DME, which... In other words, introducing the DME does not
seem to solve the problem it addresses. If I misunderstand what the
DME is supposed to be, then I'm quite happy to retract all I have said.
But I have indeed read your postings, and refrained from comment for
lack of time. I simply didn't want silence to be taken as acceptance
when you made that an issue.

Martin