[From Bill Powers (980728.0851 MDT)]
Bob C.(980727.2052 PT)--
1. Is feeling displeasure/pleasure hypothesized to be proportional to
intrinsic error as experienced by the reorganiztion system/consciousness?
I don't think the reorganizing system itself necessarily involves
consciousness. I can imagine reorganizing unconsciously. My present
hypothesis is that consciousness focusses on error in the hierarchy, and
that reorganization follows consciousness (in location, with reorganization
itself resulting from intrinsic error). The reason for this is to get
reorganization to work at the places in the learned hierarchy where
something isn't working right (as shown by abnormally large error signals),
instead of just reorganizing everything, good systems as well as bad. I
consider this strictly a conjectural hypothesis, supported only by
anecdotal data.
What we experience consciously, I have proposed, consists entirely of
afferent signals in the behavioral hierarchy that is constructed by the
intrinsic reorganizing system. The signals that the reorganizing system
detects are not the same signals we find in the hierarchy, and (I assume)
are not accessible to consciousness. What we learn to experience as hunger,
for example, is a sensory side-effect of processes like stomach
contractions that result when the real controlled variable, the
concentration of glucose in the bloodstream, falls below its intrinsic
reference level. We don't consciously experience the glucose concentration.
I suppose there could be cases where the same physiological variable is
sensed by both the reorganizing system and the
learned hierarchy, but that's not the general case I assume.
On the top of page 194 it says, "The reorganization sytem is totally
pragmatic:if it feels good, do it; if it feels bad, change." and goes on
to say "Push the change button - it's the only one there is. If the result
isn't "good", push it again."
This is speaking figuratively, as if from the point of view of the
reorganizing system. But I do not think the reorganizing system actually
thinks or has a conscious point of view. I think of the reorganizing system
as a dumb built-in device which doesn't participate in the operation of the
learned hierarchy directly.
2. On pg. 198 in talking about Wiener;s diagram of adaptive control, it says
"Unspoken in Weiner's diagram, by the way, are the reorganizing sytem's
intrinsic reference levels. They are built into the box called the
comparator
and constitute the definition of the desired characateristics of the system."
Are these intrinsic reference signals in the general sense, or just some kind
of reference signals which check if the feedback will be stable and negative?
Anything that the designer wants to use as a criterion for adaptation. The
box Weiner showed as a comparator for this system would actuually have to
perform a lot of functions beside simple comparison. Reference signals
can't do any checking; they're just signals representing the magnitude of
something. The "checking" has to be done by some circuit designed for the
purpose.
By stable and negative are you refering to just control systems in the
hierarchy or the overall loop including both the hierarchy and intrinsic
system.
Just the hierarchy. The intrinsic system is inherited, not adaptive.
I initially assumed you mean some kind of references which can be
compared to the result of the test signal to determine if the feedback is
stable and negative in the hierarchy since it does not seem possible
according
to the diagrams in the chapter that the result of the test signal will give
any info. about whether intrinsic error (intrinsic reference in general minus
intrinsic state) has been reduced by the test stimulus.
That's correct.
However, on page 200,
second paragraph, in talking about awareness it says "It merely
experiences in
a mute and contentless way, judging everything with respect to intrinsic
reference levels, not learned goals." Also, above this on pg. 200 you talk
about the purpose of volition as being "to see what will happen", and I
assume
that this is "what will happen with respect to intrinsic state."
I'm very unsure about all this stuff. Awareness seems to be like pure
observation. Volition seems to be like arbitrarily experimentation with
changing reference signals. How all this works and what it accomplishes is
only vaguely defined in my mind.
3. Related to the above question, on pg. 199, third paragraph, it says,
"Received info. is used for purposes irrelevant to the meaning (external
counterpart, that is) of the perceptual signals." Do you mean by
purposes, to
see if the feedback is stable and negative or intrinsic error reducing or
both?
I mentioned negative feedbck and stability only to introduce the
possibility that the reorganizing system might monitor variables more
complex than just blood glucose concentration or pH or body temperature. A
"purpose" in PCT is specified by a reference signal, and "the purpose" of a
control system is to make the perception match the reference signal. To
measure something like stability, the intrinsic system would first have to
receive a perceptual signal that is dependent on stability. That is, some
signal that would perhaps indicate the amount of steady-state oscillation,
or the decay time of such oscillations after a disturbance. This would
require a rather complex perceptual function doing dynamic calculations (in
the case of a "test signal" it might be a synchronous detector looking for
in-phase responses to the test signal). But don't take any of this too
seriously; I meant only to point to a _type_ of system, not to propose
anything specific. I was just nodding in the direction of Weiner's model,
not adopting it.
4. Just to make sure I understand correctly, the three boxes in the
middle of
Weiner's figure on pg. 197, high freq, oscillator, comparitor, and high pass
filter, are supposed to be the output function of the reorganization
system as
diagramed on pg. 188. The compensator and effector are considered to be some
part of the hierarchy on which awareness is focussed. Is this correct?
No, that's trying to make too specific a connection of the Wiener diagram
to mine. Wiener's three-part circuit is a whole control system in itself,
of a special design. It injects a test signal into some part of the main
control system, and detects information somewhere else in the main control
system through a filter. The test signal has a frequency that's high enough
that it has no effect on the actions of the main system, which can't behave
at that high frequency. The detected result is compared with the test
signal itself, and computations are done to deduce the form of the function
that transforms the test signal into the detected result. From that
information, this adaptive system can figure out how to change parameters
in the main system to produce a desired relationship of the test signal to
the detected result, and consequently to adjust the overall behavior of the
main system. This is about 10 times more complicated than any proposal I
want to make.
5. The last sentence of the third paragraph on pg. 199 says "It may be
feasible to allow the test stimulus signals to be injected anywhere, although
I do not consider that possibility here." I assume that what you consider in
this chapter is superimposing the test stimulus on the reference signals, and
that by anywhere you mean the other parts of the control system unit of
organization like the output and input functions. Is this what you mean?
I mean _anywhere_. You could inject the test signal into the error signal,
and measure the resulting effect on the output action. But again, don't be
distracted by this discussion. An electrical engineer could think up 20
other ways to achieve adaptation. Weiner never did anything more with his
suggestion, because he wasn't an engineer. I'm not proposing a test signal
as part of my model -- just indicating that such a thing might be possible,
since Wiener suggested it. Maybe we volitionally wiggle the wheel of the
car, and judge how slippery the road is from the results of these
pseudo-steering efforts. But this needn't be a special reorganizing mode of
operation -- it could involve a learned system whose output is the gain of
the steering control loop.
6. The fourth paragraph on pg. 199 talks about two completely different
types
of perceptions. I assume by the first you mean typical perceptual signals in
the hierarchy with out awareness, and the second is the perception of
perceiving which you think is likely awareness. I'm trying to completely
understand what you mean by the perception of perceiving and how this is
different from just perception.
It's different in terms of which system is doing it.
It seems to me that higher level input
functions could be considered the perception of perceiving because they are
derived from copies of lower level perceptions.
Yes, but they are not perceptions _of the fact that you are perceiving._ A
configuration perception is derived from a set of sensation perceptions,
but it represents the presence of a configuration, not a report to the
effect that sensation perceptions are occurring.
What seems to be different
to me is that from the way you have described things, you can't have
awareness
with out volition and vice versa.
I don't know if that's true. One can observe a nice sunset without wanting
to alter it.
So awarenes is the reception of the result
of test signals sent to the hierarchy (volition), in a similar way that a
radar image is produced by receiving signals which are sent out and bounced
off objects. A second difference is a sense of comparison with intrinsic
reference which I assume is related to feeling pleasure/displeasure. Are my
conjectures on track?
Yes, for that particular case, but not as a general requirement of the
model. Awareness is the reception of perceptual signals, however they are
generated. They aren't necessarily (or even often) generated as a result of
volitional effects. You're talking about a particular case in which we
might volitionally interfere with the perceptions that are going on, by
arbitrarily wiggling reference signals. This isn't the only, or the main,
mode of operation of the system. It's just one possible mode out of many.
7. Have you writen anything more recently on reorganization/awareness to
which you can point me?
Only in the new book, which is now available (See Amazon.com).
I'm afraid that if you want real answers to these questions, you're just
going to have to do the research that I never did.
Best,
Bill P.