Chaotic Control or not

[From Dick Robertson,2004.05.08.1820CDT]

What do you guys (and gals) think of this? It's from the April 2003 issue of the U. of Chicago
Magazine, that a friend was reminded of while reading my book,

    "Deep in the Albert Merritt Billings building associate professor of
neurology John Milton steps side to side as he balances a wooden dowel upright on
his fingertip. 'I'm getting pretty good at this,' he exclaims.
    "Milton isn't clowning around during a break from directing the
Hospitals' Epilepsy Center, where he performs implants of vagal nerve stimulators in
severely affected patients. He's demonstrating his unexpected discovery about
the body's nervous system: it generates random 'noise' to handle tasks--like
balancing a dowel or standing still without falling--that require response in
less time than it takes for a signal to travel to the brain and back (100-200
milliseconds for the stick, 250-500 milliseconds for standing). 'If the nervous
system can only make a correction every 200 milliseconds,' he asks, 'what's it
doing the other 199?'
    " 'Flipping coins.'
    "In a 2002 Physical Review Letters study, Milton and Juan Cabrera, a
physicist in Venezuela, filmed a dozen people balancing the
two-and-a-half-foot-long dowel, measured the size of its wobbles, and then correlated them with

subjects' hand movements to keep it upright. About 98 percent of the movements
occurred in less than 100 milliseconds. 'That tells us something uncontrolled by
the nervous system is at work,' says Milton. He calls that something 'noise'
'The hand makes little errors in its position. It should move a certain
time in a certain direction, but it doesn't get it exactly right.' The
difference between where the fingertip should move and where it goes is measured as
noise. And the errors, he notes, actually seem to help: the stick stay balanced
only if it is on the verge of toppling over.
    "Milton isn't sure why the errors help or how noise works--at this point,
his observations exist only as a mathematical formula. But his finding is
making neuroscientists rethink their ideas about motor control. Next he'll
research 'how you get better' at tasks that require quick but precise reflexes.
His focus will be the golf swing--something Milton, with a handicap of 6, is
already pretty good at."

Sounds like the inverted pendulum to me. I think I'll try contacting him, but, Bill Powers
would probably do a better job of it, in explaining "how the errors help," if you can find the
time.

Best,

Dick R.

From [Marc Abrams (2004.05.08.2139)]

Dick,

Llinas research has found, and the paper I cited showed this, and that
is, our motor control is _not_ continuous. This has been known for over
a century by physiologists. Parkinson's disease is a manifestation of
this gone wrong.

I think you will find that 'error' (in many guises) is indeed the
'motivating' factor in our behavior.

I feel that if researchers spent more time looking at things from a
control perspective they might have some better luck in connecting the
dots. The specific mechanisms involved may be chaotic or any number of
other variations, but the overriding theme or pair of glasses
researchers must have is one of control, or negative feedback. PCT is
simply one version of the control concept. It is not the only one, there
are others, and I don't think the final answer is known yet as too what
version or more likely, which versions (i.e. combined) will ultimately
prove to be the empirical winners.

Marc

Considering how often throughout history even intelligent people have
been proved to be wrong, it is amazing that there are still people who
are convinced that the only reason anyone could possibly say something
different from what they believe is stupidity or dishonesty.

Being smart is what keeps some people from being intelligent.

Thomas Sowell

···

-----Original Message-----
From: Control Systems Group Network (CSGnet)
[mailto:CSGNET@listserv.uiuc.edu] On Behalf Of Richard Robertson
Sent: Saturday, May 08, 2004 7:21 PM
To: CSGNET@listserv.uiuc.edu
Subject: Re: Chaotic Control or not

[From Dick Robertson,2004.05.08.1820CDT]

What do you guys (and gals) think of this? It's from the
April 2003 issue of the U. of Chicago Magazine, that a friend
was reminded of while reading my book,

    "Deep in the Albert Merritt Billings building associate
professor of neurology John Milton steps side to side as he
balances a wooden dowel upright on his fingertip. 'I'm
getting pretty good at this,' he exclaims.
    "Milton isn't clowning around during a break from
directing the Hospitals' Epilepsy Center, where he performs
implants of vagal nerve stimulators in severely affected
patients. He's demonstrating his unexpected discovery about
the body's nervous system: it generates random 'noise' to
handle tasks--like balancing a dowel or standing still
without falling--that require response in less time than it
takes for a signal to travel to the brain and back (100-200
milliseconds for the stick, 250-500 milliseconds for
standing). 'If the nervous system can only make a correction
every 200 milliseconds,' he asks, 'what's it doing the other 199?'
    " 'Flipping coins.'
    "In a 2002 Physical Review Letters study, Milton and Juan
Cabrera, a physicist in Venezuela, filmed a dozen people
balancing the two-and-a-half-foot-long dowel, measured the
size of its wobbles, and then correlated them with

subjects' hand movements to keep it upright. About 98
percent of the movements occurred in less than 100
milliseconds. 'That tells us something uncontrolled by the
nervous system is at work,' says Milton. He calls that
something 'noise' 'The hand makes little errors in its
position. It should move a certain time in a certain
direction, but it doesn't get it exactly right.' The
difference between where the fingertip should move and where
it goes is measured as noise. And the errors, he notes,
actually seem to help: the stick stay balanced only if it is
on the verge of toppling over.
    "Milton isn't sure why the errors help or how noise
works--at this point, his observations exist only as a
mathematical formula. But his finding is making
neuroscientists rethink their ideas about motor control.
Next he'll research 'how you get better' at tasks that
require quick but precise reflexes. His focus will be the
golf swing--something Milton, with a handicap of 6, is
already pretty good at."

Sounds like the inverted pendulum to me. I think I'll try
contacting him, but, Bill Powers would probably do a better
job of it, in explaining "how the errors help," if you can
find the time.

Best,

Dick R.

[From Bill Powers (2004.05.09.0533 MST)]

Dick Robertson,2004.05.08.1820CDT--

What do you guys (and gals) think of this? It's from the April 2003 issue
of the U. of Chicago Magazine, that a friend was reminded of while reading
my book,

In the spirit of maintaining a professional and substantive kind of
discussion on CSGnet, I will comment only that I disagree with almost every
detail of John Milton's facts, observations, and reasoning..

Note that in our analysis of tracking behavior, matching the model to real
behavior requires giving it a time delay of 120 to 150 milliseconds (not
250 to 500), and that despite this delay, the control system is perfectly
stable and deviations of the real behavior from the behavior of the model
show only a minor amount of unpredictable variation -- less than 2% of the
range of the variables. And the model, of course, requires no random
fluctuations at all in order to control properly.

Balancing a stick (see my demonstration of balancing an inverted pendulum)
also does not require any random variations in order to work properly. The
random variations are internally generated and make control harder, not easier.

Best,

Bill P.

From [Marc Abrams (2004.05.09.0825)]

[From Bill Powers (2004.05.09.0533 MST)]

In the spirit of maintaining a professional and substantive
kind of discussion on CSGnet, I will comment only that I
disagree with almost every detail of John Milton's facts,
observations, and reasoning..

I can understand a disagreement of reasoning. But observations and
facts?

One of the 'problems' any theorist faces is the ambiguity often involved
in observation and experimentation. As Bruce G. has pointed out, we all
tend to see things in the framework we each walk around with. This has
been a big short coming on CSGnet in that people refuse to believe that
others can actually observe the same things and come up with different
explanations then they can for the same 'observed' data.

I think an approach that showed why your observations make more sense
than theirs would be a more effective approach than telling someone that
they are 'wrong' and don't have a clue as too what they are actually
observing.

Do you really believe this statement above will get anyone who does
believe Milton to be accurate to accept your conclusions? I'm not
suggesting Milton is accurate. I'm suggesting that people who do, and
might believe Milton will not be swayed by the words you used.

Telling people that they are 'wrong' is a nasty habit that will lose you
many friends. Telling people that there are alternative ways of viewing
things might grab their interest if you present your argument with
_their_ views in mind. That is, along with telling them why they are
mistaken, to try and explain how the data could in fact produce these
different 'observations'. When you invalidate a persons beliefs you
invalidate the person and people will not put up with that. No matter
how great your theory might be.

Take this from someone who has experienced this first hand on CSGnet.

Marc

Considering how often throughout history even intelligent people have
been proved to be wrong, it is amazing that there are still people who
are convinced that the only reason anyone could possibly say something
different from what they believe is stupidity or dishonesty.

Being smart is what keeps some people from being intelligent.

Thomas Sowell

···

Note that in our analysis of tracking behavior, matching the
model to real behavior requires giving it a time delay of 120
to 150 milliseconds (not 250 to 500), and that despite this
delay, the control system is perfectly stable and deviations
of the real behavior from the behavior of the model show only
a minor amount of unpredictable variation -- less than 2% of
the range of the variables. And the model, of course,
requires no random fluctuations at all in order to control properly.

Balancing a stick (see my demonstration of balancing an
inverted pendulum) also does not require any random
variations in order to work properly. The random variations
are internally generated and make control harder, not easier.

Best,

Bill P.