Where do we go...?

CSGnet Archive CSG1993
1

From Greg Williams (930507)

Tom Bourbon (930506.1332)

Control as fact; PCT as model of fundamentals of control; HPCT as
...? I begin to see your position on this. Now where do we go? What
does it take to convince the inquisitive, but skeptical, physiologist
that Rick's spreadsheet model and the Powers-Williams arm model --
both of which are examples of modeling with HPCT -- do what none of
the alternatives do (so far as I know), and that is to behave, in
something under a day and a half (I exaggerate, but some other models
take quite a while to move, if they will move at all). How do we
convince this person to *demand* unequivocal demonstrations that the
various non-hierarchical alternatives to HPCT can behave the way their
proponents say they will? This physiologist should insist on those
demonstrations and, when they are not forthcoming, should say, "No
thanks."

Having HPCT models which behave won't be very impressive to the
skeptic who will wonder about how much of their performance depends on
the SPECIFIC arrangement of control loops postulated by Bill (higher-
level errors altering lower-level reference signals). Alternatives to
HPCT (some perhaps hierarchical themselves, but not structured exactly
as Bill Powers postulates) exist which DO behave to various extents
(the Beer bug behaviors appear a lot closer to reality than do the
Rick spreadsheet behaviors, in my opinion, and there are lots of
models in BEHAVIORAL CYBERNETICS with rich behaviors). But there are
no alternative models which attempt to do the SAME kinds of things
which Rick's spreadsheet or the arm model do. So I think what is
needed is a RANGE of models for ONE kind of behavior, to see if HPCT
can do the job "better" (match the data of real organisms) better than
the alternatives. I think that the skeptical physiologist confronted
with ONE model which matches (more or less) behavioral reality should
insist that (to begin to make choices about which model is "best")
EITHER HPCT models be made for behaviors currently modeled
in nonHPCT ways (Bill began doing this for the Beer bug but didn't
finish; the arm model approaches this, relative to some models in
BIOLOGICAL CYBERNETICS) OR nonHPCT models be made for behaviors
currently modeled in HPCT ways. Somebody (PCTer or nonPCTer) has to
make a model to confront the other side's model head-on. The arm model
is the closest PCTers have come to date.

Do you think the spreadsheet and arm models offer any
support for speculations that PCT can be expanded to HPCT?

I do, but the question is whether a skeptical physiologist would count
these as more than examples of HPCT NOT failing, with the question of
what other kinds of models ALSO might not fail left open.

Perhaps what we need to do is issue a challenge for an open competition to
model a particular kind of behavior?

As ever,

Greg

2

From Tom Bourbon (930507.0940)

From Greg Williams (930507)

Tom Bourbon (930506.1332)

Control as fact; PCT as model of fundamentals of control; HPCT as
...? I begin to see your position on this. Now where do we go? What
does it take to convince the inquisitive, but skeptical, physiologist
that Rick's spreadsheet model and the Powers-Williams arm model --
both of which are examples of modeling with HPCT -- do what none of
the alternatives do (so far as I know), and that is to behave, in
something under a day and a half (I exaggerate, but some other models
take quite a while to move, if they will move at all). How do we
convince this person to *demand* unequivocal demonstrations that the
various non-hierarchical alternatives to HPCT can behave the way their
proponents say they will? This physiologist should insist on those
demonstrations and, when they are not forthcoming, should say, "No
thanks."

Having HPCT models which behave won't be very impressive to the
skeptic who will wonder about how much of their performance depends on
the SPECIFIC arrangement of control loops postulated by Bill (higher-
level errors altering lower-level reference signals).

I know what you are saying, and it is my also point: in their training, most
behavioral and life scientists (including most of us who try to do PCT
modeling) were not prepared for the next step, which you discuss below.

Alternatives to
HPCT (some perhaps hierarchical themselves, but not structured exactly
as Bill Powers postulates) exist which DO behave to various extents
(the Beer bug behaviors appear a lot closer to reality than do the
Rick spreadsheet behaviors, in my opinion, ...

(Deja vu) "Close to reality?", he asked, with a sense of foreboding. This
is what I mean. The focus of our training was usually on outward
appearances and the focus of modeling was on making the outward appearances
"right," no matter the internal organization that was required. PCT
modelers are trying to work from the inside, to test a model of organization
that behaves. I know this sounds like a trivial distinction, or no
distinction at all, but that is not the case. See below ...

and there are lots of
models in BEHAVIORAL CYBERNETICS with rich behaviors). But there are
no alternative models which attempt to do the SAME kinds of things
which Rick's spreadsheet or the arm model do. So I think what is
needed is a RANGE of models for ONE kind of behavior, to see if HPCT
can do the job "better" (match the data of real organisms) better than
the alternatives.

I agree that this is part of what is needed, but the other possibility is
what interests me the most -- ONE model that produces a RANGE of "behaviors"
that it probably does not even know are there. One model that, with no
substantive changes in its organization (a reference signal or two, a few
gains, etc.), does at least a "pretty good job" of duplicating the results
of a lot of special purpose models, none of which generalize to a wide
range of applications. The verdict is still out on whether PCT, or HPCT, is
that kind of general model, but I have not yet seen another that is close.
The several (extraordinarily clever) special-purpose "bug" models and robots
notwithstanding.

I think that the skeptical physiologist confronted
with ONE model which matches (more or less) behavioral reality should
insist that (to begin to make choices about which model is "best")
EITHER HPCT models be made for behaviors currently modeled
in nonHPCT ways (Bill began doing this for the Beer bug but didn't
finish; the arm model approaches this, relative to some models in
BIOLOGICAL CYBERNETICS) OR nonHPCT models be made for behaviors
currently modeled in HPCT ways. Somebody (PCTer or nonPCTer) has to
make a model to confront the other side's model head-on. The arm model
is the closest PCTers have come to date.

Agreed, on the role of the arm model, but on the original issue, you are
saying that the way to convince our skeptical physiologist is for us to do
everything ourselves. This person is not much of a skeptic, if some of the
ideas I mentioned above have not come to mind.

Do you think the spreadsheet and arm models offer any
support for speculations that PCT can be expanded to HPCT?

I do, but the question is whether a skeptical physiologist would count
these as more than examples of HPCT NOT failing, with the question of
what other kinds of models ALSO might not fail left open.

That is part of what we ask of this physiologist -- a realization that the
HPCT model does not fail this test -- that is, after all, the only positive
result that can come from testing a model -- it did not fail this test, now,
on to the next one. But as portrayed so far, our physiologist seems to
demand more of HPCT, and less of other models. If so, our problem
transcends the the issue of the quality of the evidence we offer concerning
the phenomenon of control and the performance of the HPCT model.

Perhaps what we need to do is issue a challenge for an open competition to
model a particular kind of behavior?

Andy Papanicolaou has yearned for access to just one person, at one
foundation, who would come to her or his senses and realize that this is the
only reasonable approach to funding basic research and theorizing on "motor
control" -- we just need to be sure the person lets someone into the
competition who models it as perceptual control, instead!

Off to see my daughter graduate from university. Can't wait to get back on
Sunday and see what has come in on the net.
Until later,
  Tom Bourbon