(Gavin Ritz
2010.08.08.22.33NZT)
[From Bill
Powers (2010.08.08.0145MDT)]
Gavin Ritz 2010.08.08.16.03NZT –
BP: I think we need one output function per control system.
That’s exactly what I said in the last thread. For this to work at the
higher levels I think it’s a unique output structure for a unique error
signal for a unique input signal for a unique perceptual input structure all
relating to a particular unique “criteria”
You’re limiting your concept of signals too much. A signal is a variable
that can change over a range from zero to some maximum value.
I’ve got no problem with this.
If we think in terms of
balanced pairs of signals, as in my oculomotor model from the 1980s, the signal
can vary from a maximum negative to a maximum positive value.
So it’s
“unique” only in terms of identity:
This is specifically what I’m
talking about, identity. But this is s slippery subject identity as it has a
fixed like quality, but changes.
it is emitted by a
particular comparator, or balanced pair of comparators, in a particular control
system. And don’t say “relating to” unless you’re prepared to say how
the relationship is brought about.
I’m specifically talking about a
specific citeria which has a specific identity which may have different
intensities.
Also, don’t forget that we are, like it or not, simplifying the system. Those
diagrams you found unhelpful in B:CP show how what is actually a network, a
many-to-many relationship, can be represented as an equivalent set of
many-to-one functions for purposes of modeling. And finally, each perceptual
signal is a different function of overlapping sets of lower-order perceptions.
A change in one lower-order perceptual signal can alter many higher-order
perceptions of which it is a component. The perceptual signals are in
physically distinct channels, but they are related to each other by their
common ancestry at lower levels.
This is built like the number system
1+1+1+1+1……………………………………………………………………
the only difference is sequence is not requisite at all. You can have
one’s coming in from anywhere. Stuart Kauffman at Santa Fe modeled this type of structure with his network of buttons. (See
“At Home in the Universe” the search for self-organization).
So as per PCT systems concept level, only its all systems concept level. Not
really a hierarchy. This way it solves some major problems and then there is
some math one can use for this. Topoi Logic and Network Systems.
Don’t give up on the hierarchy so easily. The fact that you understand only a
few levels doesn’t mean there aren’t more. All you have to do is ask
“why” a given control system pursues a particular goal. That leads to
a higher level of control. And so on. You really can’t get away with only two
levels. In the spinal cord, the tension in tendons is controlled by a control
system in which the spinal motor neurons are the comparators and muscle fibers
are the output function.
I really do not know enough about the
details of the human neurological system to really make an educated guess. So I
guess I leave this for now.
That level of control is
used by a second level which is the stretch reflex, a combined static and
dynamic controller. That’s two levels and we haven’t even reached the brain
stem. If you lump these levels together, you’re only doing it in your head.
They remain separate levels in the organism.
Yip I see your point.
Also
similar “unique criteria” just bunch together. Just like in the
human body, there is no real level of hierarchy just different functional
systems all working together.
So what I’m saying it’s possible that there really is only two
levels, the one connected directly to the actual modalities (input functions)
and output functions (muscles) the rest is a network (all with each unique
control system), with unique signatures.
Hey just a wondering
mind!!!
No way. that’s just lazy thinking. You can’t solve the problem by pretending
it’s simpler than it is. The trick is in finding just enough simplification so
you can handle the concepts, but not so much as to conceal the real problems
that have to be solved.
The output function converts
the error signal to an output signal, perhaps by integrating it or
through some gain function, and then
it is fanned out to lower-order reference inputs.
Don’t follow this structure.
OK, I won’t follow it in this post. But see part 3 of the Byte articles.
Each lower-order
comparator can receive multiple inputs, one from each of several different
higher-order systems – summed, probably. Each higher-order system’s output can
be sent with different weightings to many lower-order comparators
Not so sure about this, I think the brain is just
billions of unique firing neural structures all linked together, some how.
“Some how” is just arm-waving. It’s like, just the whole thing, you
know. The whole question is “HOW?” Modeling is about trying to answer
that question. If you don’t even want to try, you might as well give up,.
I don’t mean this IS how
the system is hooked up, only that this is the way I’ve been thinking of it.
Okay I’ve been thinking about this a lot
too.
By the way, I apologize for underestimating your mathematical
abilities.
No worries, it’s not that big a deal
really.
Best,
Bill P.
···