From Greg Williams (920330)
EPISTEMOLOGY (and another topic, if there's room, but it's looking like there
won't be) will comprise the subject matter in the next CLOSED LOOP, due out
this month, by yours truly's unilateral decree. Those who don't like
unilaterality (and even those who do) are welcome to suggest possible subjects
for the July and later issues.
Bill Powers (920330.0800)
Some solutions to the trajectory problem came together this morning. The
model needs at least one more level, the transition level, which I now
am convinced could also be called the "path" level.
I suspect that you can get it to work, even though it isn't as elegant as pure
end-point control.
All this is very suggestive. To me, it suggests that we need one more
level, a transition level, to control the parameter that moves the
fingertip reference position from the initial position to the target
position, along a straight line in perceptual space.
Sounds like a good rule to begin with. Go for it!
I can see now that we will be able to make a prediction: the velocity profile,
normalized as in the article, will also be the same for different spatial
separations of initial and final positions, at least for rapid movements.
When confirmed, that will get some attention!
I think this addition will make the model more acceptable to
conventional modelers, because it does allow for trajectory planning.
Right on!
Greg, how much of this do you think we actually need to get working
before we submit a paper on this model?
I don't know about the "we" business -- I'm just the guy who found the
equations of motion, remember? -- but the choice seems pretty clear: end-point
control vs. path control, and I still don't think the referees are going to
like the shapes of the end-point-control trajectories. On the other hand, if
you were up front about the "limitations" of end-point control and talk about
how well path control would work, maybe they'd accept the model as it stands
now. In suppose it depends greatly on the backgrounds of the referees. Don't
wait on my account.
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.
You have deduced that cockroach movement will affect the influence of the air-
puff on the hairs. But is the loop continuously closed? You have NOT deduced
that the movement of the cockroach is influenced by the changes in influence
of the air-puff on the hairs during the movement. That is the part of the
feedback loop (if it is a loop) inside the organism.
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.
Yes, we KNOW that there is an overall feedback loop. But is a particular
cockroach escape action the result of an internal pre-calibrated (by evolution
and possibly learning) chain mechanism, or a continuous control-of-perception?
That is the $64 question. As I posted to Rick recently, in the case of a chain
mechanism, all I think is compromised with respect to PCT ideas is the notion
of control always being CONTINUOUS.
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.
I'll be in Lexington tomorrow to do research for HortIdeas; maybe I'll have
time to look for evidence one way or another in Camhi's book and elsewhere.
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 is a potentially good reason why evolution might "just" say "get away":
the neural overhead associated with PCT-circuitry might be more "costly" than
than associated with chain-circuitry. But I'd need to see the particulars
before claiming that any of these reasons, construed generally, is "good
enough."
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.
That WOULD be a giant leap. The less-giant leap which actually seems to be the
one taken by several nonPCTers is that actions (I think that is what you
meant, not PCT-type behavior, which would make the "giant leap" simply
impossible by definition!) DO immediately influence the "stimuli," but often
that influence doesn't affect the trajectory (taken generally) of the action.
The real worth of the HPCT (sic) model is in the proposition that
strong negative feedback loops are the foundation of all organized
behavior.
Fine by me. The "hangup," as I see it, is the insistence that control be
continuous.
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.
If a pre-calibrated chain mechanism contributed to higher-level control-of-
perception, it wouldn't be resisted. And if there were no reference levels at
the level of the chain mechanism, it wouldn't act like a disturbance.
Are we to give up looking for closed loops just because we can see a situation
as open-loop?
No, we should be looking for higher-level loops. And setting aside the claim
that all control is continuous.
Greg