[From Bill Powers (920330.0930)]
Greg Williams (920329) --
Are you claiming that there is actual data showing that (a) in the case
of short-duration air puffs, the body orientation changes before the >air
puff is over, and/or that (b) if the body orientation does change >before
the (say, somewhat longer-duration) air puff is over, the >movement of the
cockroach is actually influenced by the change in >alignment between the
body and the continuing air puff?
I'm making a deduction. Beer said that the turning movement was completed
in 60 milliseconds. A movement of air caused by a large approaching
clodhopper would without doubt last a lot longer than 0.06 sec. So it's
dead cert that the movement of air around the sensing hairs is affected
(strongly) by the turning movement before the movement of air has ceased,
and even before it has peaked. Actually I would expect such a surge of air
to propagate outward for several seconds, the cockroach not only turning
while the air movement is still going on, but moving downwind a
considerable distance, creating local effects that probably exceed the
amplitude of the original air puff. There can be no question that the
behavior of the cockroach strongly influences and even cancels the effects
the air puff would have if the cockroach stayed in one place. So this is a
negative feedback situation, however the cockroach is organized to behave
in it.
Because nobody has done anything like a quantitative experiment with this
escape response (I'll bet), nobody knows whether the cockroach is acting as
a control system or just goes through a fixed repertoire of actions. The
only reasons to reject the closed-loop hypothesis, so far, go like this:
feedback is too slow; evolution just says "get out of this place;" the
response (escaping) doesn't affect the stimulus (stamping your foot). Are
those reasons good enough for you?
There ARE a lot of cases where g(r) is demonstrably nonzero. But it is
quite a leap to your general claim. I wish you could see that the worth
of HCPT modelling does not hang on the truth of that generality.
No, the giant leap is the proposal that there are any cases in which
behavior does not immediately influence the stimuli that actually "caused"
it. The real worth of the HPCT (sic) model is in the proposition that
strong negative feedback loops are the foundation of all organized
behavior. If it were true that only a few systems here and there showed
this interesting property, then PCT would be just a curiosity and there
would be no reason to re-examine any S-R formulation that seems to work (no
matter how poorly). And HPCT would be nonsense, because you can't have a
hierarchy of control that includes a lot of S-R systems that behave without
regard to the rest of the system. They would just be disturbances, and the
closed-loop systems would resist their actions.
None of this means that we have to close our minds against the possibility
that some output is caused by some input that isn't in a feedback relation
to that output. Sure, it can happen. But so what? Are we to give up looking
for closed loops just because we can see a situation as open-loop? Like the
foot-stamping causing the escape response with no feedback effect on the
foot-stamping? If there really are straight-through responses, the
methodology of control theory will find them. Feedback-controlled variables
are not hypothetical; they're testable. If those apparent straight-through
responses result merely from taking too anthropocentric and general a view
of the situation, S-R theory and its methods will NEVER lead to discovery
of the truth. There is no point in assuming S-R connections, and there is
every reason to assume closed-loop connections until the data prove
otherwise. This is true even in the most solid-appearing examples of open-
loop behavior.
This has very little to do with humility on my part (perhaps for good
reason, like hubris). It has a great deal with seeing how poorly behavior
has been observed in the past, and with what giant leaps of faith the S-R
model has been defended.
ยทยทยท
--------------------------------------------------------------------
Martin Taylor (920329.1330) --
I'm not making an explanatory model of anything when I refer to the
maintenance of entropy. I'm describing negative feedback. >Environmental
disturbance increases the entropy of the system, and >negative feedback
reduces it. In a stable controlled system the >entropy remains stable
despite disturbance.
If entropy is controlled, then it's sensed, compared with a reference
entropy, and maintained at that reference level by variations in action. I
don't buy it. I don't even think that entropy is systematically affected by
negative feedback control of any perceptual variable. If you hold your arm
out straight, the entropy of your system increases more and more rapidly as
fatigue sets in; you then have to rest in order to absorb or free from
storage some negative entropy to make up for the drain.
Entropy, like information or probability, is a calculation by an observer,
not a system variable. It changes as dQ/Q, if I remember right. This
calculation can be applied to any measure of anything. It's controlled only
if the behaving system specifically senses it (calculates this function of
its inputs) and acts to keep it at some preferred level. An observer's
calculations of the entropy of some other system have no necessary
relevance to what makes that system work. If the calculations prove
consistent, then all that has been proven is that there's a consistent
side-effect of the action of the system. This does not make the side-effect
either causal or explanatory of the system's operation.
I just don't buy explanations of organized systems that relay on abstract
measures. Organized systems result from real interactions among real
variables, not from abstract characterizations of those interactions.
Yes, I mean that disturbances, within limits, will be resisted in such >a
way that the disturbed structure returns to its undisturbed state.
This is still too qualitative a statement. Does the disturbance cause any
signficant departure from the undisturbed state? Can you cite an example of
what you're talking about? Even a pendulum would fit your statement, but
it's certainly not a negative feedback control system. All the restoring
energy, as Gary Cziko mentioned, comes from the disturbance, in the
pendulum.
I maintain that the distinction you are looking for is not the high->gain
negative feedback, but the reference-signal + comparator >structure.
I'm not looking for a general abstract distinction, but for a description
of the way organisms work. High gain negative feedback (with or without a
variable reference signal, with or without an explicit comparator) creates
the kind of behavior we see in living systems at all levels from
biochemistry to control of system concepts.
What I hope to do, probably starting when I get back from my trip in
June, is to come to a resolution of the problems of the thermodynamics >of
the control system hierarchy.
Fine.
... a subjective lag of zero is not the same as a loop delay of zero.
The latter refers to how long it takes the perceiving system to acquire
information about the effect of action in the presence of all sorts of
environmental disturbances and disturbances induced by competing ECSs.
Agreed. There is, however, one objective criterion that must be met:
whatever the actual loop delay, something in the loop must insert averaging
of such an amount that the system would behave no differently if the actual
lag were zero. Otherwise the loop can't be stable.
Why do you say there are no "internal" loops? Do you mean that there >are
no loops within a single ECS?
Yes -- no behavioral loops. Any individual function (input, comparison,
output) could contain feedback loops as part of the computational process,
but the ECS as a whole has no internal loops.
Do you mean that imagination doesn't happen?
No. Imagination results from outputs of an ECS that enter an imagination
connection external to the ECS and arrive back at its input as if a real
perception were being received. This connection is external to the ECS as I
define it: input, comparison, output.
Or that kinaesthesis doesn't happen?
Kinaesthesis happens. The output of the spinal ECS affects muscles in its
environment. Those muscles have physical effects on local tissues, where
effects of independent disturbances also can occur. The result is
excitation of sensory nerves at the inputs of the ECS. I place the boundary
of a higher organism's behavioral systems so that everything outside the
nervous system is in the environment of the behaving system.
Or are you making the assertion, that I expect to take issue with >later,
that there are no cross-connections within a level of the >hierarchy?
I make no use of such cross connections but I don't deny their existence.
Some connections that appear to be cross connections are really composite
control systems -- for example, the connections that make a stretch
response on one side of a joint relax the muscle on the other side. The
opposing muscles are part of a single control system. The only true cross
connections, I think, in the terms you mean them, would be between systems
at the same level which are otherwise independent of each other. Such
connections exist but I have only a vague idea of what they accomplish.
The chaotic behaviour (or, more probably, near critical behaviour) is
required for the perceptual functions, and it is to make that behaviour
useful that we have categories. I think that's what Freeman is saying,
as well. But even if he isn't, I am.
I think even perceptual functions can become organized through random, not
chaotic, change. Freeman's concept of chaos in perception is that there are
modes of operation of a whole chunk of the brain, like the olfactory lobe,
and that perception is the existence in this chunk of some mode of large-
scale distributed oscillation. In the absence of inputs, this large system
goes into a chaotic mode, which can then settle into new basins of
attraction when novel input combinations occur, thus leading to a new mode
of perception.
While Freeman's waving his arms and conducting this music, the composition
sounds inspiring and beautifully constructed and the words to the song are
very plausible. But after the conclusion, the questions rush in. What good
does it to to have the whole olfactory system oscillating? How does that
lead to following a scent to its source? What is it that discriminates one
mode of oscillation from another, saying "That's chocolate" and "That's
perfume"? Where's the recognizer? Whqt does all this have to do with
perception? Somehow Freeman has managed to show that a large-scale mode of
oscillation can depend on inputs, but he still hasn't answered the question
of how we distinguish one input from another. I think he does some very
intensive interpreting, assuming, and even imagining when he reports the
"facts."
But I have to admit that there's a kernel of an idea there that's
attractive. One wants to believe that somehow our perceptual worlds are
truly basically alike. If there are mathematical laws that say our
perceptions naturally settle into certain preferred forms, rather than just
being scattered at random among the possibilities, then there are grounds
for thinking our perceptual worlds are pretty similar. I guess I just have
to wait on the sidelines for further (and more convincing) developments.
------------------------------------------------------------
Best to all
Bill P.