WTP & DME - RKC

<Bob Clark (940219.1750 EST)>

Bill Powers (940204.1110 MST)

It strikes me that this name is too much in the spirit of explaining
a set of functions directly in terms of what they accomplish. How
do you explain decision-making? It's done by a Decision-Making
Entity. What are the properties of a Decision-Making Entity? It
makes decisions. This does not illuminate the process of making
decisions, but only names it.

I offered the name, "Decision Making Entity," as a label for a
concept composed of a set of events I perceive within myself. I find
others reporting similar sets of events within themselves. The name
was never conceived as more than a name for a concept -- not
"explaining" anything. The name is a convenient label for a concept
that is important to the operation of adaptive control systems.

To me this is much like our selection (I think you originated it) of
the term "Feedback Function" for the concept of a mechanism that
converted external physical events into internal neural signals. It
might have been given a different name, perhaps the "Sensory
Converting Entity," or the "Input Transducer." Each name has its
implications, some of which relate to its operation.

In both cases, the concept for which the term is selected must be
joined with other components of the system to become meaningful and
useful. The concept then might be used for explanations of
observations, design of devices, analysis of operations, or other
purposes. The name, itself, is no more then a convenient shorthand.

ยทยทยท

-----------------

UNPACKING

I would try to unpack the concept of decision-making to see what
operations are required, then put _those operations_ into the model.
This would permit us to ask what _other_ processes, beside
decision-making, operations like that could be used to accomplish.
This gives the model more generality.

In various previous posts, Bill, I think I have done exactly that. I
did not put it in exactly those terms, but to me, it was equivalent.
Let me try again.

I.
The primary and defining action of the DME is that of selection and
application of reference signals for one or more levels of the
hierarchy.

II.
In order for "selection" to be meaningful, the DME must have more
than one possible alternative available.

III.
The only place where alternatives for selection can be found is in
memory.

IV.
There are three criteria for selection: a) the alternative must have
some connection with the currently perceived situation; b) it must be
possible to apply that alternative to the current situation; c) that
alternative must offer a promising outcome -- whether by a simple
re-run of the remembered events, or by extrapolation from the
remembered events.

V.
Item IV requires comparisons to be made between alternatives. These
alternatives consist of sets of perceptions, some current, some
imaginary. The concept comparison between sets of perceptions is
already available in the concept of the "Comparator Function"
included in the basic Feedback Control System. In that application,
the Comparator Function is used to compare current perceptions with
corresponding reference signals.

VI.
The criteria noted in IV are applied as inputs to the DME Comparator
Each alternative is applied to one input of the DME Comparator, and
each criterion is applied to the other input. The combined error
signals that are the output signal from the DME Comparator determine
which alternative[s] may be acceptable. The alternative that passes
this test is then applied as reference signals to the specified
locations within the hierarchy. If no alternative passes, additional
alternatives are sought.

VII.
This is a greatly condensed summary of these suggestions. Further
discussion would be very welcome.

---------------------------

With reference to my "conditions required" for operation of the DME,
namely, "consciousness" or "awareness," you remark:

Any time you're consciously doing something many lower levels are
controlling their perceptions without your conscious awareness,
although you could become aware of them, usually.

followed by:

I don't think there is any control process at any level that can't
work either consciously or unconsciously. When I say "you" do this,
I mean you-the-organism, not you-the-conscious-entity.

I have no trouble with either of these statements. To me,
consciousness is only required for the operation of the DME. Perhaps
it would be better stated as "consciousness is a _property_ of the
DME." Otherwise, as you say, the lower level systems operate by
seeking to maintain the reference levels assigned by the DME.

However, as you note above, "you could become aware of them,
usually." Thus the DME is likely to monitor the continuing operation
of the lower order systems -- perhaps only as a precaution.
Monitoring lower order operations requires very little time since
attention (awareness) can be shifted very quickly from one set of
perceptions to another. This is reported in BCP, circa page 200.
Thus the driver, while looking for house numbers, need not pay much
attention to a gust of wind -- unless it has a major effect! As you
have reported, awareness can shift very rapidly from one place to
another.

You note:

I stuck my neck out in BCP to propose mechanisms for doing things
like this, which makes the proposals vulnerable to disproof.

The potential for disproof is very important. But the opposite,
"proof," may not be possible. I am not proposing to "disprove"
anything. Instead, I am trying to improve and extend the very
thorough and basic work you have presented in BCP.

Bill, I hope this post helps clarify my suggestions. Your comments
are appreciated. They lead to further discussion.

Regards, Bob Clark