<Bob Clark (940516.1130 EDT)>
It seems to be about time to review the origins of the DME concept.
In B:CP, Chapter 15, Memory, Bill presents Figure 15.3, "Final form
of the unit of behavioral organization." This includes two single
pole double throw switches.
The perceptual switch serves to select inputs to higher order
systems. These inputs come from either lower order perceptual
signals or from a memory connection of the same order. Bill remarks,
B:CP, p 220: "It might be nice to know what changes the position of
the perceptual switch -- I have no idea at present."
The memory switch serves to select outputs from that same memory.
These outputs go either to the comparator of the same order, or to a
memory connection leading to higher order systems. After discussion
of the implications of the possible combinations of these two
switches, Bill comments B:CP, p 224: "... 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."
In the discussion of the four possible combinations of these two
switches, B:CP pp 220-222: The Control Mode; The Passive Observation
Mode; The Automatic Mode; and the Imagination mode, it becomes
apparent that the control of these two switches is critically
important.
···
-----------------
The question of the control of these switches has four possible
answers: a) the reorganizing system; b) pre-existing higher level
systems, such as established programs; c) no control at all; d) an
entity to be studied and analyzed.
a) The concept of the reorganizing system is very powerful, but it is
limited to strictly random actions, and acts only when there is
noticeable intrinsic error.
In ordinary situations, there is minimal Intrinsic Error, so the
reorganizing system, with its random processes, is not involved.
Thus the changes in behavior resulting from operation of the
switch-controller do not involve reorganization.
b) Combinations of switches selected by an existing program can
provide the necessary control. But it is necessary to select the
program to be implemented. How can this be done? Also, if no
suitable program is available, how can a new program be created?
c) Without some form of "switch controller," the switches remain
unchanging. Therefore the various high level reference signals are
also unchanging. This is a purely mechanical set of systems. To an
observer, it is an S-O-R system, with a complex form of "O." To the
theorist, the observer's "stimulus" becomes a "disturbance." And the
observer's "response," becomes the system's "output." Given the
initial state of the system, its performance is completely
predictable.
d) I have suggested naming the "switch controller" the Decision
Making Entity. This is simply a label, with minimal connotations, to
facilitate further investigation of its properties, etc. Originally,
the DME was no more than that.
However further consideration suggested some characteristics that
seem necessary. For example, the individual must be conscious --
otherwise the systems simply continue to do whatever they were doing.
Also, the DME must have some basis for selecting the positions of the
switches, and some way to operate them. Continuing this analysis has
led to my previous suggestions.
--------
A complete theory of behavior should be equally applicable to the
theorist, the experimenter, and the observer. If these people have
no way to control the two switches, they, too, are unable to explore
new ideas, design new experiments, or investigate new situations.
There may well be better ways to handle this situation than those I
have suggested. But the resolution of these questions is imperative.
Regards, Bob Clark