Reorganization

CSGnet Archive CSG2010
41

Bill Powers (2000.08.08.0923 MDT)–

Doing anything that requires a lot of effort entails large error
signals, because error signals are all that drive actions.

Bill Powers (2000.08.09.0857 MDT)–

normal conditions involve forces that are
only a small fraction of the maximum force possible, so the error
signals required to produce them are also relative small.

What is wrong with the following reasoning?

If a large error is easy to reduce to near zero the effort to do so is not great. Conversely, if a small error is difficult to reduce to near zero (and gain is high enough – how near zero I want it), the effort to do so can be great. The strength of effort is related to the strength of disturbances. Yes, error signals are all that drive actions; but no, an action does not require a lot of effort solely in proportion to the magnitude of the error signal, but also depending upon what resistance it meets.

Strictly, effort refers to output functions in intensity-level control loops. Suppose I am correcting a small error in my perception of the position of a door but the effort is great because the door is stuck. If the output is maxed out and the reference calls for an even higher level of output, then the error signal at the intensity level is high. If the reference level is high, but the intensity of effort is able to match it (output is not maxed out), then the error signal might be near zero. So long as the door is stuck, there is an error at the higher level (door position), and that’s why the intensity reference is high.

Bruce gregory (2000.0808.1308)–

I’m flabbergasted. As long as control is maintained, why should

error signals be large? Clearly I fail to understand something very

fundamental. When I pedal my bike, I assumed that my actions were

driven by changing reference levels rather than by large errors.

IF your actions were driven by changing reference levels, your bike ride

would be very unpleasant and, most likely, very painful.

You’re both right. As the reference value departs from the current perceptual input the error signal increases. Also, as the perceptual input departs from the current reference level, due to disturbances, the error signal increases. It doesn’t matter how they come to differ. Both happen as I ride my bike.

···

At 04:02 PM 08/08/2000 -0400, Lazare, Mark Crisis counselor, Phoenix AZ wrote:

42

In a message dated 8/9/2000 10:22:55 AM US Mountain Standard Time,
powers_w@FRONTIER.NET writes:

<<
>IF your actions were driven by changing reference levels, your bike ride
>would be very unpleasant and, most likely, very painful.

Huh? To push down on a pedal, you raise the reference signal for the system
that controls sensed pedal pressure. To turn left, you set the reference
signal for the bank angle of the bicycle to a small leftward lean, which is
produced by setting the reference signal for the wheel angle so the wheel
cocks, for a moment, to the right ... and on on. It's reference signals all
the way down.

Best,

Bill P.
  >>
I think we are talking about different levels in the HPCT.

I was commenting on the "bike ride" -- Which is a Task equlivant to the 7th
level in your Living Control Systems 1989.

The things you metion above happen at lower levels and are not prof you are
riding a bike.
1) you could be on a exersie bike in the living room.
2) you could be flat on your back in a exercise class "pedeling"

When you learn to ride a bike you might focus on the Motor output, above a
bit -- but you soon learn to the 2 most important things are to keep it in
motion and don't fall off. Those are the references one is controlling.
Ballance and Motion. I am still riding a bike, just coast'n down hill. I
would not Change my referance for ballance in the middle of my bike ride, I
would surely fall off.

I will compensate with my vert behavioor, pedal more going up hill, swerving
to miss cars and pot holes. My refereces in riding a bike are Keeping it in
motion and Dont fall off. I could be doing the same behavior with equivlant
effort, but I am in my living room on a exersise bike and not going anywhere.
"To Ride a Bike" is the g

43

I Sent my last E-mail a before I was done with it. (hit the wrong button -- I
guess I was still partly asleep)

<<<Bill: Huh? To push down on a pedal, you raise the reference signal for
the system that controls sensed pedal pressure. To turn left, you set the
reference
signal for the bank angle of the bicycle to a small leftward lean, which is
produced by setting the reference signal for the wheel angle so the wheel
cocks, for a moment, to the right ... and on on. It's reference signals all
the way down.>>>
Bill----
I was commenting on the "bike ride" -- Which is a Task equlivant to the 7th
level in your Living Control Systems 1989.

The things you metion above happen at lower levels and are not proof you are
riding a bike.
1) you could be on a exersie bike in the living room.
2) you could be flat on your back in a exercise class "pedeling"

When you learn to ride a bike you might focus a bit on the Motor output,
above -- but you soon learn to the 2 most important things are to keep it in
motion and don't fall off. Those are the references one is controlling.
Ballance and Motion. I am still "riding a bike" just coast'n down hill. I
would not Change my referance for ballance in the middle of my bike ride, I
would surely fall off.

<<>>

"To Ride a Bike" is the goal -- you either do it or not. Not doing it to me
means, not getting on the bike or falling off or having a wreck on it. All
the behavior related to riding the bike is compensatory to all the possible
disturbance of riding a bike (maintaing motion and ballance). Disturbances
of Gravity, Grade of the road, pot holes, on coming traffic, rabid dogs,
water puddles and so on.

Mark

44

[From Bruce Gregory (2000.0811.1006)}

8/11/00 10:53:11 AM, "Lazare, Mark Crisis counselor, Phoenix AZ"

The things you metion above happen at lower levels and are not

proof you are

riding a bike.

When did the goal become to prove you are riding a bike?

1) you could be on a exersie bike in the living room.
2) you could be flat on your back in a exercise class "pedeling"

When you learn to ride a bike you might focus a bit on the Motor

output,

above -- but you soon learn to the 2 most important things are to

keep it in

motion and don't fall off.

Remind me never to go bike riding with you :wink:

Those are the references one is controlling.

Ballance and Motion. I am still "riding a bike" just coast'n down

hill. I

would not Change my referance for ballance in the middle of my

bike ride, I

would surely fall off.

You have a fairly unsophisticated notion of balance.

<<>>

"To Ride a Bike" is the goal -- you either do it or not. Not doing

it to me

means, not getting on the bike or falling off or having a wreck on

it. All

the behavior related to riding the bike is compensatory to all the

possible

disturbance of riding a bike (maintaing motion and ballance).

Disturbances

of Gravity, Grade of the road, pot holes, on coming traffic, rabid

dogs,

water puddles and so on.

Good luck!

BG

···

<DTSDTO@aol.com> wrote:

45

[From Bill Powers (2000.09.20.0208 MDT)]
The shuttle landed safely a few minutes ago.

Bruce Abbott (2000.09.19.1150 EST)--

Here's an article I thought might be of interest to CSGnetters:

Corbetta: Understanding how babies' brains work

Taking two steps forward sometimes requires a step back for babies acquiring
new motor skills, according to onging research at the Purdue Infant Motor
Development Laboratory.

I don't suppose that these researchers cited Frans and Hedwig Plooij's much
more extensive studies of infants, in which they showed that at many points
in development where a new level of skill was about to be acquired, the
behavior of the infants became less organized and regressed to an earlier
stage.

Best,

Bill P.

46

[From Bruce Abbott (2000.09.20.1955 EST)]

Bill Powers (2000.09.20.0208 MDT)]

I don't suppose that these researchers cited Frans and Hedwig Plooij's much
more extensive studies of infants, in which they showed that at many points
in development where a new level of skill was about to be acquired, the
behavior of the infants became less organized and regressed to an earlier
stage.

This was a newspaper article about Corbetta's research, not a scientific
report, and as such did not contain references.

I don't know of any requirement implied in HPCT that individuals regress
during reorganization. If lower levels of control remained stable while the
next higher levels were being formed, rather than the observed "regression,"
this also would be consistent with the HPCT model. Correct?

Bruce A.

47

[From Bill Powers (2000.09.21.0655 MDT)]

Bruce Abbott (2000.09.20.1955 EST)]

I don't know of any requirement implied in HPCT that individuals regress
during reorganization. If lower levels of control remained stable while the
next higher levels were being formed, rather than the observed "regression,"
this also would be consistent with the HPCT model. Correct?

I think what happens is that when reorganization starts working at a higher
rate, some existing higher-order systems become _dis_organized, leaving
lower-level or cruder systems (i.e., typical of an earlier stage of
organization) as the only ones working properly. That's what gives the
impression of regression. Remember that e. coli-type reorganization does
not head directly toward an improved state of affairs: it's a random walk.
You expect "progress" in the wrong direction a significant part of the
time, under this model.

Best,

Bill P.

48

[From Bruce Abbott (2000.09.21.1745 EST)]

Bill Powers (2000.09.21.0655 MDT) --

Bruce Abbott (2000.09.20.1955 EST)

I don't know of any requirement implied in HPCT that individuals regress
during reorganization. If lower levels of control remained stable while the
next higher levels were being formed, rather than the observed "regression,"
this also would be consistent with the HPCT model. Correct?

I think what happens is that when reorganization starts working at a higher
rate, some existing higher-order systems become _dis_organized, leaving
lower-level or cruder systems (i.e., typical of an earlier stage of
organization) as the only ones working properly. That's what gives the
impression of regression. Remember that e. coli-type reorganization does
not head directly toward an improved state of affairs: it's a random walk.
You expect "progress" in the wrong direction a significant part of the
time, under this model.

O.K., thanks for the clarification. Given the data, it appears that some
existing systems must be replaced or modified to make them compatible with
the new mode of locomotion. This sounds to me more like reorganization of
existing systems at some level of perceptual control as opposed to the
building of a higher level of perceptual control. In the latter case, it
might not be necessary to alter anything at the existing level, in which
case "regression" would not be expected.

Bruce A.

49

[From Bill Powers (2000.09.22.1441 MDT)]

Bruce Abbott (2000.09.21.1745 EST)]

Given the data, it appears that some
existing systems must be replaced or modified to make them compatible with
the new mode of locomotion. This sounds to me more like reorganization of
existing systems at some level of perceptual control as opposed to the
building of a higher level of perceptual control.

I think both happen. When a higher level reorganizes, or becomes organized
for the first time, existing lower-level systems may have to be altered to
work together as required by a higher system. For example, control
processes which have been used only one at a time may become unstable when
they are both used at once to accomplish a new higher-level task that
requires their coordination.

On a less theoretical level, even people who know how the
four-buttons-and-lights task works have to reorganize a little when the
connections between buttons and lights are randomly shuffled for a new run.
Simply changing from waiting for the next light to come on to anticipating
it (once the new sequence is memorized) takes a moment to get used to, and
performance deteriorates as mistakes are made.

The Plooijs, of course, were studying infants learning new kinds of control
processes for the very first time. A lot more could be done toward studying
this phenomenon.

In the latter case, it
might not be necessary to alter anything at the existing level, in which
case "regression" would not be expected.

The term "regression" is metaphorical: it looks as if the organism has gone
back to an earlier stage of development. But that's only an appearance,
resulting from temporary loss of some skill with the result that earlier
levels or kinds of skills, which are not as good at correcting errors, are
the only ones left operating. You see this in the childish behavior of
drunks whose higher levels of organization are impaired. Athletes appear to
experience something like this, too, when they overtrain or become
overanxious about success. They temporarily lose the peak of performance
and have to wait until it comes back. I think this is particularly visible
in the game of golf, where even the best golfers have their "A" days and
their "B" days. Golf requires essentially impossible levels of control of
body motions, and if reorganization continues after maximum performance has
been reached -- for the moment -- the result is deterioration again. If
you're too calm, you don't do enough reorganizing, and if you get too
excited or worried about doing well you reorganize right past the peak.
I've been noticing this for a few years now, and think there's something to
study in that idea.

Best,

Bill P.

50

[From Bill Powers (2000.09.22.1441 MDT)]

Golf requires essentially impossible levels of control of

body motions, and if reorganization continues after maximum performance

has

been reached -- for the moment -- the result is deterioration again. If
you're too calm, you don't do enough reorganizing, and if you get too
excited or worried about doing well you reorganize right past the peak.
I've been noticing this for a few years now, and think there's something

to

study in that idea.

Bill, if you can use PCT to help the golfers lower their scores you'll
finally achieve that elusive goal: PCT will become THE MODEL for
psychology!! Go for it!

David

51

[From Bruce Abbott (2000.09.19.1150 EST)]

Here's an article I thought might be of interest to CSGnetters:

Corbetta: Understanding how babies' brains work

Taking two steps forward sometimes requires a step back for babies acquiring
new motor skills, according to onging research at the Purdue Infant Motor
Development Laboratory.

"These findings indicate that as a baby learns a new skill, such as walking,
the brain appears to reorganize itself, resulting in a temporary
developmental step backward," says Daniela Corbetta, assistant professor of
health, kinesiology and leisure studies and lab director.

Corbetta's findings follow her observations of hand preference and skills
such as crawling and walking.

"It has been well-documented since the 1940s that infants return to
two-handed reaching and grabbing for a period of time even though
single-handed reaching has been mastered for several months. But until now
there has never been a documented explanation for why it happens," she says.

Corbetta set out to identify at least one explanation by observing nine
babies beginning at seven and eight months of age until they started to walk
on their own.

"We presented each baby with a series of small, colorful objects at the
midline, and all of them were reaching for it with one hand until they began
to walk," Corbetta says. "Once they started walking, they all reverted to
bimanual reaching for a period of time."

She got the same results when the infants retrieved a small toy from a box.

Before they could walk, the babies consistently opened the box with one
hand; but after they began walking, they switched back to two hands, she
says. "So this was not just happening with one task, it was happening
across all the tasks we gave them."

Corbetta surmises that learning to walk requires a great deal of postural
reorganization, as many muscles are used in new ways. How the brain
processes walking alters old reaching tendencies temporarily.

···

--------------------------------------------------------------------------
[_Inside Purdue_, Sept 6, 2000, p. 7]

There's a nice picture included showing Corbetta testing one of the infants
(Collin). In part the caption says "Collin has been coming to the Purdue
Infant Motor Development Laboratory since he was six months old so Corbetta
can record changes in his reaching patterns during both the onset of
crawling and the onset of walking." Collin is shown seated at a small desk,
opening the lid of a small box with one hand and reaching for the object
inside with the other, while Corbetta holds the bottom portion of box from
the rear sides.

Bruce A.

52

This is Phil Runkel querying n 19 Oct at 11:10 am PST.

If control takes longer at the higher levels, then presumably
reorganization does so, also, since reorganization is a search for
control. So maybe dreaming is a part of reorganization at the higher
levels. Maybe I should say conscious levels.

53

[From Bill Powers (2003.10.19.1453 MDT)]

···

At 12:13 PM 10/19/03, you wrote:

This is Phil Runkel querying n 19 Oct at 11:10 am PST.

If control takes longer at the higher levels, then presumably
reorganization does so, also, since reorganization is a search for
control. So maybe dreaming is a part of reorganization at the higher
levels. Maybe I should say conscious levels.

Could be. Maybe what's bad about sleep deprivation is simply lack of
reorganization to take care of internal conflicts that build up every day.

Best,

Bill

54

[From Rick Marken (941127.2045)]

I'll be out of town for a week so I won't get to watch the progress of
Bruce's reorganization; but it looks like some reorganization may be
starting. For example, Bruce Abbott (941127.1900) says:

In other words, if we change the environment so that there is nothing to
learn, e. coli will learn nothing. Brilliant! Poof, the law of effect is
disproven!

Anger is a sign that reorganization may be in progress. In the hopes
of keeping you reorganizing, let me again point out that there WAS something
to learn in my "de-artifacted" version of the E. coli experiment; it
was still true that you could control the spot if you pressed when the
spot was moving away from the target and didn't press when the spot was
moving toward the target. People easily learn to do this; the control
model does this too. Only the law of effect model fails to control the
spot because the consequences of presses no longer have the artifact that
makes a reward for action more likely when the spot is moving away from the
target and less likely when it is moving toward the target.

All my "de-artifacting" code does is make a rewarding change in gradient
just as likely to follow a press when going up as when going down the
gradient. This change in the situation doesn't take away the ability to
control the spot -- any more than the change from an FR 1 to an FR 100
takes away a rat's ability to control the rate of food input. This change
prevents your law of effect model from controlling but it DOESN'T
prevent people or a control model from controlling (or learning to
control). Of course, in order to control in the "de-artifacted" situation
the subject (and model) have to press a lot more often to get the
desired consequence (becuase, for example, the conseuqnce of a press
when the spot is going away from the target is now most likely to STILL
BE the spot going away from the target). But controllers, who are not
controlled BY the consequences of their actions, simply keep on pressing
until they get the consequence they want.

Have a great week.

Rick

55

Reorganization
[From Rick Marken (2003.12.11.1540)]

I’m going through all kinds of computer changes so I don’t know if this will work.

I heard a talk yesterday by Judy Singer, who is the head of Education at Harvard, and she presented some data that might be pertinent to how reorganization might work. It’s individual data from a set of kids playing a game called Fox n’ Geese. I think it might be a game like Missionaries and Cannibals where you have to move everybody from one side of the river to another using a boat that can hold only 2 without ever having the number of bad guys (Cannibals or Foxes, I suppose) outnumber the good guys (Missionaries and Geese).

The data [which I hope shows up below as an image] shows, for each kid, the number of moves made on each trial (game) until a “catastrophic event). I think it the kids make it to 20 moves without a catastrophe they’ve solved the problem, at least one " game”. Some kids clearly never get it (never learn to control the result); two seem to get it but can’t quite keep it.

If you believe that the kids are learning a control skill via reorganization then you will see the moves up until solution as basically random. Learning theorists tend to see a gradual improvement for all the kids.

I want to find out more details about what the kids actually did in this study but the research is available only in a Harvard doctoral dissertation from 1980. I might be able to get more detail from Singer’s new book, in which these data are reported, but that will take some time.

Anyway, I just thought I would put some data out there that seems pertinent to the question of how reorganization might work.

Best

Rick

Richard Marken, Ph. D.

Senior Behavioral Scientist

RAND Corporation

1700 Main St., P.O. Box 2138

Santa Monica, CA 90407-2138

310 393-0411 Ext. 7971

rmarken@rand.org

ole0.bmp (87 Bytes)

56

[From Dick Robertson,2003.12.11.2300CST]

"Marken, Richard" wrote:

Rick, notice that this page is totally blank. This has happened to me
several times lately, yet I see later that others must have got the
message because I see their replies to it. I have asked before if any
one can explain what happened to my receipt but nobody replied. Can you
cue me itn?

Thanks, Dick R

57

[From Rick Marken (2003.12.11.2220)]

Dick Robertson (2003.12.11.2300CST) --

"Marken, Richard" wrote:

Rick, notice that this page is totally blank. This has happened to me
several times lately, yet I see later that others must have got the
message because I see their replies to it. I have asked before if any
one can explain what happened to my receipt but nobody replied. Can you
cue me itn?

I don't understand it, either. I presume you talking about the message I sent earlier today with the subject head "Reorganization". Did you not even see the text message? I sent it using Outlook on a PC. Even on my Mac I get a little icon that says ole0.bmp. I can't read the bitmap on the Mac but I can download it and see it on PC. As I said in the post, the data bore a strong resemblance to the data you got with Glines using the game Bill invented.

The PDF file that contains the data is at

http://gseacademic.harvard.edu/alda/Handouts.htm

Click on the bullet called "Discontinuous and Non-Linear Change". The data is on p. 10.

Best

Rick

···

---
Richard S. Marken
marken@mindreadings.com
Home 310 474-0313
Cell 310 729-1400

58

[From Bruce Nevin (2003.12.12 11:24 EST)]

Dick Robertson,2003.12.11.2300CST --

"Marken, Richard" wrote:

Rick, notice that this page is totally blank.

The text body of this message was formatted with HTML. Your email software has to be able to display "styled text".

It's possible for the sender to configure their email software to accommodate recipients who can view text only. If you're using Eudora, go to the Tools menu, select Options, scroll to the "Styled Text" option, and select "send both plain and styled". For an explanation, go to Help > Topics and search for the topic on "styled text".

Outlook probably works similarly. I don't use Outlook because of its vulnerabilities to viruses.

         /Bruce Nevin

···

At 10:59 PM 12/11/2003 -0600, Richard Robertson wrote:

This has happened to me
several times lately, yet I see later that others must have got the
message because I see their replies to it. I have asked before if any
one can explain what happened to my receipt but nobody replied. Can you
cue me itn?

Thanks, Dick R

59

[From Dick Robertson,2003.12.12.2330CST]
(Actually wrote this yesterday, but the students at NEIU seem to keep
the server tied up until the wee hours, so let me try this AM.)

Rick Marken wrote:

> Dick Robertson (2003.12.11.2300CST) --
>
> "Marken, Richard" wrote:
>
> Rick, notice that this page is totally blank. This has happened to
me
> several times lately, yet I see later that others must have got the
> message because I see their replies to it. I have asked before if
any
> one can explain what happened to my receipt but nobody replied. Can

> you
> cue me itn?

I don't understand it, either. I presume you talking about the message

I sent earlier today with the subject head "Reorganization". Did you
not even see the text message?

No, the page was totally blank.???

I sent it using Outlook on a PC. Even on
my Mac I get a little icon that says ole0.bmp. I can't read the bitmap

on the Mac but I can download it and see it on PC.

I get it on my PC through Netscape Communicator from the Univ. Could it
be that Netscape doesn't accept it?

As I said in the
post, the data bore a strong resemblance to the data you got with
Glines using the game Bill invented.

Interesting. I will try

  http://gseacademic.harvard.edu/alda/Handouts.htm

Click on the bullet called "Discontinuous and Non-Linear Change". The
data is on p. 10.

Well, that might look a little like our results, but the graph isn't
long enough to see where it would go. Their approach seems to make the
research unduly complex.

Thanks for your advice. I wonder how I can stop getting those blank
pages in the occasional post. It just happens once in a while.

Thanks to Jim and Bruce N. for your contributions too. See my next
posts.

Best, Dick R

60

Re: Reorganization
[Martin Taylor 2003.12.13.09.24]

[From Rick Marken
(2003.12.11.1540)]

The data [which I hope
shows up below as an image] shows, for each kid, the number of moves
made on each trial (game) until a “catastrophic event). I think
it the kids make it to 20 moves without a catastrophe they’ve solved
the problem, at least one " game”. Some kids clearly never
get it (never learn to control the result); two seem to get it but
can’t quite keep it.
If you believe that
the kids are learning a control skill via reorganization then you will
see the moves up until solution as basically random. Learning
theorists tend to see a gradual improvement for all the
kids.

I’m reminded of something W. P. Tanner told me about a study he
did some time before 1960. I’ve no idea whether it was published.
Maybe some old-timer from UMich around 1960 would know.

Tanner was looking for a demonstration that one could learn a
totally new perception of a kind not normally encountered, in this
case an auditory one. He presented a tonal ramp to each ear, going
from zero to some maximum level in one ear while it went from maximum
to zero in the other. The total power to the two ears stayed constant
throughout. The perceptual result would be that the tone swept from
one side of the head to the other. I think the total sweep took 100
msec, but I could be wrong on that.

Onto this sweep he added a signal in one of four possible times.
It consisted of a short interval (10 msec?) in which all the power
went to one ear, still keeping the total power in the signal constant.
The job of the subject was to say whether the blip happened in
interval 1, 2, 3 or 4. He thought that nobody would have ever
encountered this kind of signal before, let alone being asked to
discriminate such a subtle distinction in it.

At first, nobody could do this task. The subject was always told
the correct answer after each choice. They worked, I think, for about
an hour every day for as many as 40 days. Some of the subjects never
got above chance, but for most there was a day or two when they went
from near chance to being almost perfect. They had learned a
completely new kind of perception (and presumably would have been able
to control it, had they been asked).

I think the Fox and Geese game also involves learning a new
perception, but at a rather more abstract level. I’m not sure what
kind of reorganization is involved in the new perception in Tanner’s
study. But the results do seem to point in the same direction–not
perceiving and then suddenly being able to perceive the new
structure.

Martin