How to write a letter ?

[From Bill Powers (990103.1215 MST)]

Rupert Young (981229.2000 UT)--

If it's your output you're really controlling, then you ought to be able to
do this blindfolded, right?

But I do know what the _sequence_ is, blindfolded. At this level it looks
like a plan about to be executed with specific outputs. Only at the lower
levels is the requirement of feedback and control of input more obvious,

to >me at least.

Maybe so, but if you're blindfolded you don't know what element of the
sequence is currently occuring, and so don't know where you are in the
sequence. You need feedback to tell you where you are so you can know when
to do the next element.

One of the things that I am trying to get my head round is how the variables
and functions at the program level are represented in the brain. Do you
have any specific thoughts on that or do you just assume that they are
possible ?

I really don't know, except from observing how I think. I don't perceive
any of the functions, except as perceptual variables (a = f(b) is a
collection of symbols that I recognize as referring to a function, but I
don't experience the function-ness directly). At these higher levels my
ignorance far exceeds my knowledge.

Best,

Bill P.

[From Bruce Gregory (990103.1640 EDT)]

Bill Powers (990103.1215 MST)

Maybe so, but if you're blindfolded you don't know what element of the
sequence is currently occurring, and so don't know where you are in the
sequence. You need feedback to tell you where you are so you can know when
to do the next element.

Not really. You can type blindfolded. You know perfectly well where you are
in the sequence of letters you intend to produce.

Bruce Gregory

[From Rick Marken (990103.1550)]

Bill Powers (990103.1215 MST)

You need feedback to tell you where you are so you can know when
to do the next element.

Bruce Gregory (990103.1640 EDT)--

Not really. You can type blindfolded. You know perfectly well
where you are in the sequence of letters you intend to produce.

Are you presenting this as evidence that you don't need feedback
about the state of a sequence in order to be able to control it?
If so, this would be evidence that at least one kind of behavior is
_not_ the control of perception. Have you thought about how well
you could control the sequence of letters (blindfolded) on one
one of those keyboards that switches from Qwerty to Dvorak --
with someone randomly throwing the switch from one key mapping
to the other while you type?

Best

Rick

···

--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken/

[From Bruce Gregory (990103.1905 EST)]

Rick Marken (990103.1550)

Bill Powers (990103.1215 MST)

> You need feedback to tell you where you are so you can know when
> to do the next element.

Bruce Gregory (990103.1640 EDT)--

> Not really. You can type blindfolded. You know perfectly well
> where you are in the sequence of letters you intend to produce.

Are you presenting this as evidence that you don't need feedback
about the state of a sequence in order to be able to control it?
If so, this would be evidence that at least one kind of behavior is
_not_ the control of perception.

Nothing so grandiose.

Have you thought about how well
you could control the sequence of letters (blindfolded) on one
one of those keyboards that switches from Qwerty to Dvorak --
with someone randomly throwing the switch from one key mapping
to the other while you type?

Such disturbances would, of course, destroy control. They would not destroy
my ability to "emit" a sequence, however. That was my only point. I can type
with my fingers not placed on the home keys. (My favorite example:
University of Vsligotnis, Lod Snhlrd, Vsligotnis.) The results are not what
I intended, but I controlled the sequence just fine (If you see were my
fingers were, you can read my message.)

Bruce Gregory

[From Rick Marken (990103.1650)]

Bruce Gregory (990103.1905 EST) --

Such disturbances [changed keyboard mapping] would, of course,
destroy control. They would not destroy my ability to "emit" a
sequence, however. That was my only point. I can type with my
fingers not placed on the home keys. (My favorite example:
University of Vsligotnis, Lod Snhlrd, Vsligotnis.) The results
are not what I intended, but I controlled the sequence just
fine (If you see were my fingers were, you can read my message.)

In this example, you are not emitting a sequence of outputs; you
are still controlling a sequence inputs. In this case its a sequence
of _kinesthetic_ perceptions rather than the sequence of visual
perceptions (letters) in Bill's original example. If I injected
you with a chemical that blocks the sensory signals coming from
your fingers you could not reliably "emit" the sequence of motor
signals that produces the sequence of kinesthetic perceptions in
your example. So far, there is no evidence -- including the
evidence from those grisly deafferatiation experiments -- that
organisms can _reliably_ emit (control) any output -- let alone
a sequence of outputs.

Best

Rick

···

--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken/

[From Bruce Gregory 9990103.2005 EST)]

Rick Marken (990103.1650)

In this example, you are not emitting a sequence of outputs; you
are still controlling a sequence inputs. In this case its a sequence
of _kinesthetic_ perceptions rather than the sequence of visual
perceptions (letters) in Bill's original example. If I injected
you with a chemical that blocks the sensory signals coming from
your fingers you could not reliably "emit" the sequence of motor
signals that produces the sequence of kinesthetic perceptions in
your example. So far, there is no evidence -- including the
evidence from those grisly deafferatiation experiments -- that
organisms can _reliably_ emit (control) any output -- let alone
a sequence of outputs.

This seems to be a religious debate. I've sworn off them for the new year. I
stand by my last post. The kinesthetic feedback while typing with my fingers
on the wrong keys told me when to strike the next key, but not what key to
strike. I struck the wrong keys, but the sequence was correct. If this is
impossible according to PCT please forgive me.

Bruce Gregory

[From Rick Marken (990103.1740)]

Bruce Gregory (990103.2005 EST)--

This seems to be a religious debate.

No. It's a scientific debate. The question is whether it is
possible to emit a consistent output without perceiving any
consequence of that output.

I stand by my last post. The kinesthetic feedback while typing
with my fingers on the wrong keys told me when to strike the
next key, but not what key to strike. I struck the wrong keys,
but the sequence was correct. If this is impossible according
to PCT please forgive me.

OK. What you said in your last post was:

Such disturbances would, of course, destroy control. They would
not destroy my ability to "emit" a sequence

I didn't notice the scare quotes around "emit". Clearly what
you meant was "...to an observer, it now looks like I am
emitting a sequence without feedback. But, of course, this
is not the case. I am now controlling kinesthetic feedback
perceptions that are invisiable to the observer".

I thought you were arguing that perceptual feedback was not
necessary to produce a sequence. But now I see that you
were _not_ making that argument. You were just pointing out that
you could control the sequence blindfolded by controlling a
different perceptual variable (kinesthetic rather than visual).
Good point!

Best

Rick

···

--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken/

[From Bill Powers (9901094.1020 MST)]

Bruce Gregory (990103.1640 EDT)--

Maybe so, but if you're blindfolded you don't know what element of the
sequence is currently occurring, and so don't know where you are in the
sequence. You need feedback to tell you where you are so you can know when
to do the next element.

Not really. You can type blindfolded. You know perfectly well where you are
in the sequence of letters you intend to produce.

Yes, and that's because you're perceiving what you're doing at the sequence
level, isn't it?

Best,

Bill P.

[From Bruce Gregory (990104.1240 EST)]

Bill Powers (9901094.1020 MST)

Bruce Gregory (990103.1640 EDT)--

>> Maybe so, but if you're blindfolded you don't know what
element of the
>> sequence is currently occurring, and so don't know where
you are in the
>> sequence. You need feedback to tell you where you are so
you can know when
>> to do the next element.
>
>Not really. You can type blindfolded. You know perfectly
well where you are
>in the sequence of letters you intend to produce.

Yes, and that's because you're perceiving what you're doing
at the sequence
level, isn't it?

I guess I don't understand why blindfolding you prevents you from
"perceiving at the sequence level".

Bruce Gregory

[From Bill Powers (990105.0936 MST)]

Bruce Gregory (990104.1240 EST)--

>Not really. You can type blindfolded. You know perfectly
well where you are
>in the sequence of letters you intend to produce.

Yes, and that's because you're perceiving what you're doing
at the sequence
level, isn't it?

I guess I don't understand why blindfolding you prevents you from
"perceiving at the sequence level".

It prevents you from knowing what element of the intended sequence is
actually present, as opposed to the element you desire to be present. You
illustrated the problem yourself, the typing example with your left hand
out of position by one letter. The letter-sequence you intended came out
wrong because you weren't looking at the letter that actually resulted, and
thus didn't correct the sequence error (LOD instead of LOS). Your fingers
moved in the right sequence because you were _feeling_ which finger was
used and in what position, but the wrong _letter_ sequence occured because
you weren't _looking_ at the result. Of course if your fingers had happened
to stay in the right relationship to the keys, you would have produced the
right letter sequence, but that proves nothing -- you can't show what is
under control unless there are appropriate disturbances.

Best,

Bill P.

[From Rupert Young (981223.1710 UT)]

I've been giving some thought to how control systems at different levels fit
together with respect to the variables being controlled and the references
being set, as it's not clear to me. I thought it would be a good idea to go
through an everyday task that involves many levels and try to identify what
levels are involved and what is being controlled at each level and what is
output. Writing a letter seems like a good example and here's my attempt. As
always, please let me know where I'm going astray.

1. The reference perception I am controlling is "write a letter" which will
exist (or the error will be non-negative) until the letter is complete in
front of me. To start with it seems we are at the program level when we want
to define the _type_ of letter. So we will have a program saying something
like "if business letter then sequence A, if personal letter then sequence B",
where the sequences define the different parts of the letter like address
positions, date, signature etc. The output will be a sequence that I want to
perceive at the next (lower) level. What is the input to this (program) level ?

2. So I'm sitting at my desk with pen in hand and paper oriented correctly in
front of me. I'm writing a business letter so the I set as a reference at the
sequence level sequence A (my address top right, date, their address top left,
opening, body of letter, signature). I can see why it seems natural to talk
about control of output as it does seem sensible to think of this sequence as
what is about to be my output.

3. But first I have to position my pen so I control a relationship between the
tip of my pen and a point on the paper a couple of inches in from the right at
the top. The perception being the difference, visually, between pen and
point. The output is movement of the hand. When they coincide no further
action is taken.

4. I am controlling "my address top right", the first in sequence A, but, is
it an event or another sequence ? Can we have sequences within sequences ?
Well it seems like a sequence as it involves writing name, street, city,
county etc. Also, my name is a yet another sequence, of letters. Anyway,
controlling a sequence involves setting the references of lower levels in a
certain order. So to write my first name I control the sequence of letters
R-u-p-e-r-t first setting a lower reference of "R", the other letters wont be
set as references until the preceding one is complete.

5. The letter "R" is a reference at the configuration level (or is it event).
To write the letter "R" requires controlling the position of the pen
(relationship), the speed of movement of the pen (transition), the feel of how
tightly the pen is held (sensation) and the sensed forces of the fingers
(intensity).

6. I write the letter "R" in one go starting bottom left. I hold the pen
firmly, but softly, applying enough (but not too much) pressure on the paper
such that it is easy to move the pen across the paper and that the shade of
ink is to my liking. I move the pen across the paper in a swift movement that
allows the tip to constantly change position resulting in the letter "R"
appearing. Is this a transition level system ? Does this mean that the
output of the transition level (the x, y reference pen position of the lower
level) is
changing resulting in the "R" ? How is it changing ?

7. Anyway, we continue on in this manner until the entire letter is written.
I guess it is useful to bear in mind that we are talking about a massively
parallel set of massively connected neural control systems such that many
different variables at many different levels are being controlled
_simultaneously_. So while I am controlling the variable related to the
pressure of the pen on the paper I am also controlling a variable related the
sequence that makes up my address as well as the orientation of my head
related to the paper, the dryness/wetness of my throat, the light falling on
my retinas, the letter writing program, and many, many more....

I think I'll leave it there and wait for feedback as it's getting quite
confusing and complex.

Comments ? Marks out of 10 ? Waste-paper bin ? Actually, what would be useful
would be a diagram of all the control systems involved, indicating all the
levels, perceptions and outputs. Anyone free over the next few days ?

Regards,
Rupert

[From Bill Powers (981223.1405 MST)]

1. The reference perception I am controlling is "write a letter" which will
exist (or the error will be non-negative) until the letter is complete in
front of me. To start with it seems we are at the program level when we want
to define the _type_ of letter. So we will have a program saying something
like "if business letter then sequence A, if personal letter then sequence

B",

where the sequences define the different parts of the letter like address
positions, date, signature etc. The output will be a sequence that I want to
perceive at the next (lower) level. What is the input to this (program)

level ?

I suppose the reference program is selected to satisfy a principle: "Reply
in the appropriate format?".

2. So I'm sitting at my desk with pen in hand and paper oriented correctly in
front of me. I'm writing a business letter so the I set as a reference at the
sequence level sequence A (my address top right, date, their address top

left,

opening, body of letter, signature). I can see why it seems natural to talk
about control of output as it does seem sensible to think of this sequence as
what is about to be my output.

If it's your output you're really controlling, then you ought to be able to
do this blindfolded, right?

7. Anyway, we continue on in this manner until the entire letter is written.
I guess it is useful to bear in mind that we are talking about a massively
parallel set of massively connected neural control systems such that many
different variables at many different levels are being controlled
_simultaneously_. So while I am controlling the variable related to the
pressure of the pen on the paper I am also controlling a variable related the
sequence that makes up my address as well as the orientation of my head
related to the paper, the dryness/wetness of my throat, the light falling on
my retinas, the letter writing program, and many, many more....

Yes, it gets complicated -- and it takes a lot longer to describe than to
do it. One thing that helps is to actually do the process instead of making
it a thought experiment. Thought experiments aren't very useful, because
you leave out details that are actually there and insert details that don't
actually occur. Many things that happen are not controlled -- they just go
along with controlling the variables you do control. Example: reaching out
for a glass of water. Are you controlling the orientation and position of
the back of your hand? You could say yes, because the back of your hand
moves while you're reaching. But if you actually reach for something, I
think you'll decide that where the back of your hand is is of no
importance. This shows why it's important to deal with actual cases, not
made-up ones.

Incidentally, the reference condition is more like "a letter has been
written" rather than "write a letter." It describes the states of your
perception when there is zero error, no?

Best,

Bill P.

[From Bruce Gregory (981223.1710 EDT)]

Bill Powers (981223.1405 MST)

Rupert:

>I can see why it seems natural to talk
>about control of output as it does seem sensible to think of this

sequence as

>what is about to be my output.

Bill:

If it's your output you're really controlling, then you ought
to be able to
do this blindfolded, right?

I think Rupert's point is that it "seems natural to talk about control
of output," not that what _is_ involved is control of output. The word
control is really being used in two ways. The everyday "control of
output" and the technical "control of perception." One might even say
that one controls (everyday) one's output by controlling (PCT) one's
input. I don't think Rupert is confused, but I suspect that more than a
little confusion about PCT stems from these two quite different meanings
of control.

Bruce Gregory

[From Fred Nickols (981225.1317)]--

Rupert Young (981223.1710 UT)

Rupert offers up writing a letter as a case in point
and then inquires after the levels involved. I will
follow the responses because I'm also interested in
some everyday examples of PCT in action. And, I'll
offer up a comment or two of my own.

I would never have thought to describe the task in
the analytical way Rupert approached it. I would
have treated it as something similar to the outfielder
catching a fly ball, that is, as simply sitting down
and writing (and perhaps rewriting) until the letter
was as I wanted it. In short, I would have closed
the gap between the letter I was writing and the
letter I wanted written.

Now, right about here is where things get interesting
because I am convinced that we often are unable to
articulate in advance the result we're after. Instead,
clarity emerges from a cycle of acting and reflecting
on the effects of our actions. In other words, I don't
start out with a letter (or an e-mail) in the conscious
forefront of my mind; I simply start writing. At some
point I declare myself done. (I'm sure someone will
remark, "Yeah, and it shows in your writing, too.")

The way I say this in other circles is to say that
clarity emerges more often than it pre-exists. What
this leads me to ask of this list (and I apologize if
it's been thoroughly discussed before) is this question:

"Is it reasonable to view reference conditions as dynamic,
that is, as emerging and changing, as "snapshots" of what
is really a state of flux instead of as static?"

I ask because it seems to me that much of what I call
"configured" work isn't work carried out according to any
predefined plan or set of standards but is instead work
that is performed in concert with a developing set of
standards or reference conditions.

In Rupert's case, it is not likely, then, that I have a
hard and fast set of reference conditions for a specific
letter (although perhaps I might have some criteria for
a general class of letters such as business letters or
personal letters and so on).

To recap: I write; when I am finished, my letter meets
my criteria (i.e., is consistent with my reference
conditions). However, these criteria or reference
conditions emerged in the course of writing the letter
instead of starting out as a set of parameters imposed
on the writing task up front.

Aaarrgh! Is this communicating anything at all?

···

--

Regards,

Fred Nickols
Distance Consulting
http://home.att.net/~nickols/distance.htm
nickols@worldnet.att.net
(609) 490-0095

[From Bruce Gregory (981225.2110 EDT)]

Fred Nickols (981225.1317)

Aaarrgh! Is this communicating anything at all?

Of course. In the vernacular, "How can I know what I think, until I hear
what I say?"

Bruce Gregory

[From Bill Powers (981226.0708 MST)]

Fred Nickols (981225.1317)]--

Rupert offers up writing a letter as a case in point
and then inquires after the levels involved. ...

Now, right about here is where things get interesting
because I am convinced that we often are unable to
articulate in advance the result we're after. Instead,
clarity emerges from a cycle of acting and reflecting
on the effects of our actions.

I agree that this often happens, especially when we are learning to control
something for the first time. In many control processes, the reference
signal defines only a few variables in a situation that may involve changes
in dozens of variables. Yet all these variables affect each other, so
changing the ones we intend to change results in changes in all the rest,
too -- like the unavoidable motion of your elbow when you pull something
toward you. Your elbow might threaten to knock something off the table, so
you have to modify your actions to keep from creating other errors. I think
a lot of the "emergence" you're talking about comes from these interactions
amoung control systems. I spent yesterday with three grandchildren, two of
whom are still in the stage of knocking things off the table while pulling
a toy toward them, so I have a recent and vivid picture of this in mind.

But you're talking about something else that doesn't necessarily go away
with practice. When I write a letter (or a post) I may start with only a
general notion of "replying", but what constitutes a reply develops only as
I write. What this says to me is that most dimensions of the letter are
uncontrolled -- they take whatever form is needed to create a reply in the
particular limited sense I had in mind. If I accidentally delete the post
and have to rewrite it, I may generate the same "reply" but it is very
unlikely to come out in the same words, or even the same sequence of
points. The things that vary during the re-write are just the means of
control, so they can vary without my correcting the changes. They're not
under control.

There. I've now satisfied the reference condition I had in mind -- that
last paragraph means (to me) what I wanted to mean. And that last
_sentence_, although you can't see it, was edited to change "say" to
"mean", because as I read it over I realized it sounded as if the _words_ I
was "saying" were the point, whereas it is the _meaning_ I want to get across.

This shows that as we write, what we read, if it doesn't mean exactly what
we want it to, consistutes a disturbance, and we have to edit what is
already written to correct the error. I do a large amount of correcting
such errors before I finally send a post. That process is invisible to the
reader, of course, as is going back to correct typos (originally spelled
"tyoped").

All this is in line with your comment,

The way I say this in other circles is to say that
clarity emerges more often than it pre-exists. What
this leads me to ask of this list (and I apologize if
it's been thoroughly discussed before) is this question:

You then ask

"Is it reasonable to view reference conditions as dynamic,
that is, as emerging and changing, as "snapshots" of what
is really a state of flux instead of as static?"

Well, yes, because all reference signals are dynamic, being the outputs of
higher systems. What seems to me to happen during writing is that the
reference conditions for the words and letters I type and retype are highly
dynamic, while the reference conditions for the sentences and paragraphs
are much less so, although they still change as I reread or reflect on what
I have written so far and see just how its meaning compares with the
meanings I'm trying to communicate. Usually there's some relatively fixed
idea I have pretty clearly in mind, and most of the variations occur as I
see the flaws in my attempt to communicate it. But even that isn't always
true -- I may, for example, finally get what I mean down on paper, and then
realize that I don't want to get into _that_ subject with _this_ person at
all, and delete the whole thing including the desire to communicate that
meaning.

I think we can say that clarity emerges because we start with a clear idea
about something at a high nonverbal level of perception, and then, as we
work out the details, see the meaning of what we are writing gradually
approach the intended meaning.

I ask because it seems to me that much of what I call
"configured" work isn't work carried out according to any
predefined plan or set of standards but is instead work
that is performed in concert with a developing set of
standards or reference conditions.

This is probably true, but the changing reference conditions are being
changed BY SOMETHING, and that, I would say, is a higher-level system
trying to control for its own reference condition -- "clarity", perhaps. As
recent events in Washington have shown, such a system can control for a low
degree of clarity as well as a high degree.

Aaarrgh! Is this communicating anything at all?

I agree. That is the highest level of reference condition in writing
letters, or posts.

Best,

Bill P.

[From Rupert Young (981229.2000 UT)]

Bill Powers (981223.1405 MST)]

If it's your output you're really controlling, then you ought to be able to
do this blindfolded, right?

But I do know what the _sequence_ is, blindfolded. At this level it looks
like a plan about to be executed with specific outputs. Only at the lower
levels is the requirement of feedback and control of input more obvious, to me
at least.

Incidentally, the reference condition is more like "a letter has been
written" rather than "write a letter." It describes the states of your
perception when there is zero error, no?

Yes, I was wondering about that.

One of the things that I am trying to get my head round is how the variables
and functions at the program level are represented in the brain. Do you have
any specific thoughts on that or do you just assume that they are possible ?

Regards,
Rupert

[From Rick Marken (981230.0900)]

Rupert Young (981229.2000 UT) --

One of the things that I am trying to get my head round is how
the variables and functions at the program level are represented
in the brain. Do you have any specific thoughts on that or do
you just assume that they are possible ?

Bill seems to be away so let me give this a preliminary shot.
I think the basic assumption in PCT is that controlled variables
are represented as perceptual signals that vary only in magnitude.
So a program level controlled variable is represented in the
brain in exactly the same way as an intensity level variable;
as an afferent neural signal that varies in magnitude (rate of
firing).

Here's a VERY simple schematic of a program level perceotual
signal:

              -->|layers of | -->|program level|
environmental -->|perceptual| -->|perceptual |--->program perceptual
variables -->|processing| -->|function | signal

The magnitude of the "program perceptual signal" represents the
degree to which a particular program (the one being "looked for"
by the "program level perceptual function") is happening in the
environment. The reference signal that specifies the desired
magnitude of the "program perceptual signal" is also just a
neural signal of a particular magnitude; the magnitude of the
reference signal, therefore, specifies the degree to which the
program being looked for by the "program level perceptual function"
should occur. It seems likely that at the program level the
perceptual signal ("program perceptual signal") is binary; the
program is either happening or it's not. So the reference would
only have two meaningful states; 1 (the program is happening) or
0 (the program is not happening).

A perceptual function like the "program level perceptual function"
is obviously a very complex neural network with time lags and
logical decision points and so on. It's more than just a
transducer (such as the ones in the sensory receptors that
convert environmental energy into rates of neural firing); it's
a very complex computation device with memory and processing
capabilities.

At least, that's the way I see it.

Best

Rick

···

--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken