Simple Simon

[From Bruce Abbott (970227.1010 EST)]

Bruce Gregory (970226.1030 EST) --

Control is the _only_ problem. Beat that for simplicity :wink:

Or naivete. Breathtaking!

It's _simple_ all right, I'll grant you that. (;->

But seriously, if what you mean by that is that all other questions about
behavior and mental processes are in a sense subsidiary to the problem of
control, I agree. All those other systems in the brain are there in the
service of control.

Simply yours,

Simon

[From Bill Powers (970227.1707 MST)]

Bruce Abbott (970227.1010 EST)--
(writing to Bruce G.)

But seriously, if what you mean by that is that all other questions about
behavior and mental processes are in a sense subsidiary to the problem of
control, I agree. All those other systems in the brain are there in the
service of control.

The point easily gets overlooked that all we can see of human behavior are
the actions that are being used to control one perception or another. And
all we can see of perceptions (in another person) are the aspects of the
environment we assume, guess, or hypothesize are being perceived. There's
really no direct way, outside neurology and introspection, to study perception.

Consider the example you brought up, judging the straightness of a line. I
can think of at least four perceptions that would be equivalent to
perceiving straightness: (1) perceiving the ends and middle of the line and
judging the displacement of the middle relative to the ends; (2) perceiving
curvature as a local feature of the perceptual field (involving a large
number of rods and cones, and circuitry in the retina); (3) perceiving
invariance of the appearance of the line as the eyes scan along it; (4)
perceiving the lateral shift of a small portion of the line relative to the
background as the focus of the eyes tracks along the line. There are
probably more.

My point is that when a person reports "straight" or "curved," you don't
actually know what perception is being described (except in terms of your
own way of perceiving straightness or curvature).

PCT gives us a way of investigating perception that can isolate the
lower-order aspects of perception that define a higher-order perception.
This is done simply by giving the person control over the perception by
different kinds of means and under different circumstances. When the person
is controlling the perception in a specified state, we know we are looking
at the elements of the environment from which this perception is being
derived. This can help greatly in guessing what the person means by words
describing a perception, and even in guessing about possible mechanisms of
perception.

For instance, suppose you ask a person to keep two halves of a line straight
when it passes on a slant behind an obscuring rectangle (the carpenter's
illusion). When the person adjusts the line so it looks like one continuous
straight line, you can measure the magnitude of the illusion for all sorts
of sizes and shapes of obscuring rectagles and other figures, and as a
function of the angle between the line and the side of the rectangle.

Some perceptual studies have used this "method of adjustment," and some
haven't. We have to ask just how much real information we have obtained from
studies in which the perception was not controlled, but only described. I
contend that unless you use a controlled perception, you leave the question
of what the person is perceiving essentially undefined. Saying that a person
is perceiving "straightness" doesn't tell you much.

Rick, how about a Java program that measures the carpenter's illusion as a
function of angle, using the Test?

Best,

Bill P.

[From Bruce Gregory 9970228.1100 EST)]

Bruce Abbott (970227.1010 EST)]

But seriously, if what you mean by that is that all other questions about
behavior and mental processes are in a sense subsidiary to the problem of
control, I agree. All those other systems in the brain are there in the
service of control.

Bill Powers (970228.0520 MSt)to

Bruce Abbott

I think this is an excellent description of how a control system appears to
an external observer who doesn't know about control. Present an input -- get
an output. I'm sure you realize that this is exactly the situation in which
I recommend going back to the basic phenomenon and looking for a missing
control system. The question to ask is, "What perception is being affected
in one direction by the "input", and in the opposite direction by the "output"?"

Bruce, I think the contrast between these two statements
illuminates Rick's apparent stubborn refusal to accept much of
what you say. You _do_ believe that "All those other systems in
the brain are there in the service of control." But this belief
does not yet inform the way you look at "traditional"
psychological questions. In a sense you have too _many_ arrows
in your quiver. Rick has only one, but he uses it with deadly
effect. If you want to come out ahead, you've got to beat him
at his own game. Bill is showing you the way. If you take up
the challenge, I'll be here to cheer you on!

Please forgive the extent to which this sounds like a sermon. I
really don't want to preach at you or anyone else. My
inclination is to trash this presumptuous note, but I'm going
to send it and hope you are feeling tolerant!

Bruce Gregory

For reasons that are beyond my ken, mail sent to CSGNet from
my leonardo address takes well over 12 hours to post. For reasons
that will be obvious from the first line of the post below,
I really couldn't wait that long on this post so I am trying it
post but, hey, it's my birthday;-)

路路路

from my other address. Sorry for what will surely be a duplicate
--------------------------
[From Rick Marken (970228.0950 PCT)]

Today is my birthday so everyone has to be nice to me:-)

Bill Powers (970227.1707 MST)--

Rick, how about a Java program that measures the carpenter's
illusion as a function of angle, using the Test?

Ok. I'll put that in the queue. I haven't done any Java programming
in a couple of weeks and I've got to get back to it or I'll forget how.
But I have three Java projects on which I plan to embark (now four
with the illusion project) as soon as I get a weekend or evening when
I have some free time.

One of the projects, which I started last night, is another approach to
looking at this "open loop control" thing. The idea is very simple.
The subject tracks a sine wave target; the cursor disappears from view
for a few seconds, every few seconds, and I measure tracking error with
the cursor visible (control of perception) and with the cursor invisible
(presumably control of a modelled percpetion).

My little variant on this study is simply to have the the feedback
function (the effect of the mouse on cursor position) be either
proportional to (as is usually the case) or the time integral of mouse
position. I am thinking that the model based controller would be
expected to do equally well (when comparing performance in the invisible
to that in the visible cursor condition) in the proportional and the
integral feedback condition. The PCT model should do quite differently
IF the ability to "control" in the invisible cursor condition is based
on the subjects control of the perception of hand movements relative to
target movements (maintaining this relationship in the integral
condition would be difficult because of the required phase lag between
mouse and cursor movement).

I have already done a preliminary test of this idea and it looks like
control of an invisible cursor is a LOT worse (relative to control of
the visible cursor) with the integral than it is with the proportional
feedback function.

By the way, I know that this study is architected like a conventional
psychological experiment. The IV is the type of feedback function; the
DV is the measure of control (RMS error and/or stability). What makes
this experiment appropriate to the study of control is that we know
what variable is under control (cursor/target distance) and we can
monitor the behavior of this variable under the dfferent experimental
conditions. Also, I will not consider this experiment a "success" or
"failure" until I have been able to compare the results to a working
model of a perceptual controller and to a "modelled perception
controller" to see how these systems actually behave.

Best to all

An older (but wiser) Rick

[From Bruce Gregory (970228.1650 EST)]

For reasons that are beyond my ken, mail sent to CSGNet from
my leonardo address takes well over 12 hours to post.

I've been encountering the same 12 hour delays. I've tried not
to take them personally, but sometimes I wonder...

Rick Marken (970228.0950 PCT)

Today is my birthday so everyone has to be nice to me:-)

MANY happy returns.

Bruce Gregory

[

From Bill Powers (970228.1603 MST)]

Today is my birthday so everyone has to be nice to me:-)

You might as well give up, Rick, you'll never catch me. Anyway, Happy
Birthday, old friend.

One of the projects, which I started last night, is another approach to
looking at this "open loop control" thing. The idea is very simple.
The subject tracks a sine wave target; the cursor disappears from view
for a few seconds, every few seconds, and I measure tracking error with
the cursor visible (control of perception) and with the cursor invisible
(presumably control of a modelled percpetion).

My little variant on this study is simply to have the the feedback
function (the effect of the mouse on cursor position) be either
proportional to (as is usually the case) or the time integral of mouse
position.

AH! That should do it. That will nicely screw up using kinesthetic or visual
perceptions of the mouse to fill in for the missing visual perception, and
since that is probably how people manage to keep "tracking" at all in the
blind, I predict (as you do) really poor tracking without the visual input.
Good idea!

Best,

Bill P.

[From Rick Marken (970228.0950 PCT)]

Today is my birthday so everyone has to be nice to me:-)

Bill Powers (970227.1707 MST)--

Rick, how about a Java program that measures the carpenter's
illusion as a function of angle, using the Test?

Ok. I'll put that in the queue. I haven't done any Java programming
in a couple of weeks and I've got to get back to it or I'll forget how.
But I have three Java projects on which I plan to embark (now four
with the illusion project) as soon as I get a weekend or evening when
I have some free time.

One of the projects, which I started last night, is another approach to
looking at this "open loop control" thing. The idea is very simple.
The subject tracks a sine wave target; the cursor disappears from view
for a few seconds, every few seconds, and I measure tracking error with
the cursor visible (control of perception) and with the cursor invisible
(presumably control of a modelled percpetion).

My little variant on this study is simply to have the the feedback
function (the effect of the mouse on cursor position) be either
proportional to (as is usually the case) or the time integral of mouse
position. I am thinking that the model based controller would be
expected to do equally well (when comparing performance in the invisible
to that in the visible cursor condition) in the proportional and the
integral feedback condition. The PCT model should do quite differently
IF the ability to "control" in the invisible cursor condition is based
on the subjects control of the perception of hand movements relative to
target movements (maintaining this relationship in the integral
condition would be difficult because of the required phase lag between
mouse and cursor movement).

I have already done a preliminary test of this idea and it looks like
control of an invisible cursor is a LOT worse (relative to control of
the visible cursor) with the integral than it is with the proportional
feedback function.

By the way, I know that this study is architected like a conventional
psychological experiment. The IV is the type of feedback function; the
DV is the measure of control (RMS error and/or stability). What makes
this experiment appropriate to the study of control is that we know
what variable is under control (cursor/target distance) and we can
monitor the behavior of this variable under the dfferent experimental
conditions. Also, I will not consider this experiment a "success" or
"failure" until I have been able to compare the results to a working
model of a perceptual controller and to a "modelled perception
controller" to see how these systems actually behave.

Best to all

An older (but wiser) Rick

[From Bruce Abbott (970228.1250 EST)]

Bruce Gregory 9970228.1100 EST) --

Bill Powers (970228.0520 MSt)to

Bruce Abbott

I think this is an excellent description of how a control system appears to
an external observer who doesn't know about control. Present an input -- get
an output. I'm sure you realize that this is exactly the situation in which
I recommend going back to the basic phenomenon and looking for a missing
control system. The question to ask is, "What perception is being affected
in one direction by the "input", and in the opposite direction by the

"output"?"

Bruce, I think the contrast between these two statements
illuminates Rick's apparent stubborn refusal to accept much of
what you say. You _do_ believe that "All those other systems in
the brain are there in the service of control." But this belief
does not yet inform the way you look at "traditional"
psychological questions. In a sense you have too _many_ arrows
in your quiver. Rick has only one, but he uses it with deadly
effect. If you want to come out ahead, you've got to beat him
at his own game. Bill is showing you the way. If you take up
the challenge, I'll be here to cheer you on!

I debated whether to explicitly diagram the control system I envisioned at
work in the second phase, where the participant is just emitting judgments
based on the lines presented, and decided in the interest of brevity to just
make the following verbal statement:

The fact that she is providing these judgements shows that she is
controlling something, but it is not the length of the line and it is not
the output of the judgement function. If she is controlling for giving the
experimenter what he has asked for (her honest appraisal of the length of
the line), then her judgements and the measured lengths of the lines will
continue to agree within the limits of her accuracy in making the judgements.

Thus I am _not_ asserting that the judgement is taking place outside of a
closed loop.

Bill has suggested _another loop_ in which the judgments-system may
participate during this latter performance, namely one in which the
judgement is evaluated against what the person herself judges she "should"
be saying based on the sensory input. I imagine the person saying something
like "35 millimeters --- no, wait, 30, 32 millimeters." I can think of two
or three ways to model this as a product of control action; for example, the
input line may establish a kind of "feeling" for what the judgment should
be, which is compared to the judgement one begins to announce; if these
differ the person begins to vary the judgement output until it matches the
"feeling" for what the judgement should be.

But then you have to ask, "where does this 'feeling' come from?" My answer
is that it comes from the learned function that relates the input line
length to the judgment one "feels" is correct. If the control system just
described is doing its job, then the (final) announced judgement will
correspond to the output of the function, or perhaps vary closely around it
depending on what disturbances are acting on the system.

By adding this additional loop to our model, we get a detailed view of (one
possible organization of) the system in question, but it doesn't change my
conclusion one bit. The relationship between line length input and the
underlying judgement tendency ("feeling") is open loop, and we can learn
what that relationship is to a good approximation by observing the objective
relationship between the input line lengths and the resulting announced
judgements, so long as we can assume that the person has a reference for
doing as we ask. Thus traditional methods that assume an open-loop process
will generally yield correct functions.

Please forgive the extent to which this sounds like a sermon. I
really don't want to preach at you or anyone else. My
inclination is to trash this presumptuous note, but I'm going
to send it and hope you are feeling tolerant!

No, I don't mind at all. But think about this example. It shows that
traditional methods can produce reliable data about a system component even
though it is embeded within a larger context of control.

And just for the record, other "traditional" psychophysical methods exist
that explicitly and consciously allow the participant to control the input.

Regards,

Bruce

[From Bruce Abbott (970228.0755 EST)]

Bill Powers (970227.1707 MST) --

For instance, suppose you ask a person to keep two halves of a line straight
when it passes on a slant behind an obscuring rectangle (the carpenter's
illusion). When the person adjusts the line so it looks like one continuous
straight line, you can measure the magnitude of the illusion for all sorts
of sizes and shapes of obscuring rectagles and other figures, and as a
function of the angle between the line and the side of the rectangle.

Some perceptual studies have used this "method of adjustment," and some
haven't. We have to ask just how much real information we have obtained from
studies in which the perception was not controlled, but only described. I
contend that unless you use a controlled perception, you leave the question
of what the person is perceiving essentially undefined. Saying that a person
is perceiving "straightness" doesn't tell you much.

A familiar example of a psychophysical method that centers on control is Von
Bekesky's technique for determining the intensity thresholds for tones of
various frequencies. (The method is now a standard one in use by
audiologists to assess a person's hearing.) The person puts on a pair of
earphones and asked to push down the button on a hand-held switch until she
hears a tone. While the button is depressed, the intensity of a tone being
presented to, say, the right ear, gradually increases. When the button is
released, the intensity slowly decreases, and the person is told to press
and hold the button down the moment the tone is no longer audible.

The result is the the intensity of the delivered tone is controlled at the
person's threshold for that tone. Thresholds for different frequencies
presented either to the right or left hear are assessed in this way and used
to plot a curve of threshold as a function of frequency, for each ear.

Rick, how about a Java program that measures the carpenter's illusion as a
function of angle, using the Test?

This is also called the Poggendorff illusion after its discoverer. (Yes,
that's two g's and two f's.) The typical way it has been measured is to
allow the participant to vary the position of the second segment until it is
judged to align with the linear continuation of the first segment:

                        > > /
                        > >/ <----varied up and down
                        > >
                        > >
                        > >
             fixed---> /| |
                      / | |

The size of the illusion can then be determined from the deviation of the
line's position from the position it should actually have. A computer would
make it possible to vary the angle of two lines (disturbance) while the
person attempted to maintain the variable-position line at the apparent
extension-point of the fixed-position line.

I don't know how many remember the programs I wrote (was it last year or the
previous one?) to assess what variable was being controlled in the "inverted
T" illusion. The participant used the mouse to vary the length of the
vertical segment of the T as the length of the horizontal segment varied at
random within certain limits, attempting to keep the two lengths
perceptually the same. The resulting data (collected at the rate of 60
samples/second) were written to a file, which was then used as input to a
program that fit the data to a control model. We tried a couple of ways to
define the CV and found that one fit the data better than the other.

I addressed Bill's (970228.0520 MST) "challenge" in a reply to Bruce Gregory
this morning, but alas, I have not yet seen it appear on CSGnet, although a
much latter birthday message arrived almost immediately on posting. If it
doesn't show up by morning I'll resend it.

Regards,

Bruce

[From Bill Powers (970228.1953 MST)]

Bruce Abbott (970228.0755 EST)]

I'm pleased to learn that the "Poggendorff Illusion" is actually tested
using control methodology. I do remember your experiment with measuring the
"inverted-T" illusion that way, too. Martin Taylor has reported using this
method of adjustment to measure kinetic illusions. All of these experiments
are good PCT whether called that or not, and the results are obviously worth
preserving. It would be very useful if some PhD candidate were to go through
ALL illusions and re-measure them using this technique, and even to extend
the method to experiments with non-illusory perceptions.

Of course while one is at it, it might be nice to add disturbances and
measure the control parameters, while one is in a position to do so.

Best,

Bill P.

[From Bill Powers (970228.1826 MST)]

Bruce Abbott (970228.1250 EST)--

The fact that she is providing these judgements shows that she is
controlling something, but it is not the length of the line and it is not
the output of the judgement function. If she is controlling for giving
the experimenter what he has asked for (her honest appraisal of the
length of the line), then her judgements and the measured lengths of the
lines will continue to agree within the limits of her accuracy in making
the judgements.

Thus I am _not_ asserting that the judgement is taking place outside of a
closed loop.

Bill has suggested _another loop_ in which the judgments-system may
participate during this latter performance, namely one in which the
judgement is evaluated against what the person herself judges she "should"
be saying based on the sensory input. I imagine the person saying
something like "35 millimeters --- no, wait, 30, 32 millimeters." I can
think of two or three ways to model this as a product of control action;
for example, the input line may establish a kind of "feeling" for what the
judgment should be, which is compared to the judgement one begins to
announce; if these differ the person begins to vary the judgement output
until it matches the "feeling" for what the judgement should be.

But then you have to ask, "where does this 'feeling' come from?" My
answer is that it comes from the learned function that relates the input
line length to the judgment one "feels" is correct. If the control system
just described is doing its job, then the (final) announced judgement will
correspond to the output of the function, or perhaps vary closely around
it depending on what disturbances are acting on the system.

These ideas represent a progression toward a purely PCT view, but I can
think of ways to get still closer. When I say to you, "my keyboard is about
18 inches wide," what happens inside you (other than just reading those
words)? Do you get some notion of physical width from this? I do. If you
told me this, I could hold my hands far enough apart to give me the same
sense of width, and think "Hmm, that wide? That seems like too much."

If I'm your "talking ruler," you hear my words and convert them into spatial
perceptions, don't you? You do, that is, if the words imply something within
your experience ("The Orion Nebula is 5000 light years away" probably
doesn't fit that condition). When I say "six inches tall" you can visualize
something about that tall, which you would also judge to be "six inches
tall" if you had to act as the talking ruler.

So what is happening when a person is acting as a talking ruler? I would
guess that the person sorts through imagined verbal size-judgements until
one is found which, if someone else uttered it, would evoke a picture of
something of about the right size. I just did a quickie experiment with
Mary, holding up a letter-opener and asking how long it was. After a pause,
she said "seven inches." Later, it turned out to be 7-1/4 inches long. But I
asked her (because of the pause) if that was the first number she thought.
She said no, the first number was 6 inches. Why did you reject that, I
asked. Because it was too short, she said. It took a while to agree that the
thought "six inches" gave rise to a sense of length that was too short
compared with the sense of length obtained from looking at the
letter-opener. "7 inches" gave a sense of length that was much closer. The
object, apparently, was to perceive two senses of length and control for
their being the same. One of them came from the visual image; the other one
came from (imagined) words referring to lengths.

This implies that we learn words for units of length and can translate from
the words to the direct sense of length that accompanied them when we
learned them.

While there may be other plausible scenarios, this one fits my own
impressions and Mary's description. What we are doing when we act as talking
rulers is to select descriptions that give us a sense of length (just as if
someone else had uttered them) that is close to the sense of length we get
from the object to be measured. Since we make most of the adjustments
covertly, uttering only the final description when we're satisfied with it,
it seems to others that we simply respond to the perception of the object by
uttering words for lengths. However, if one were to avoid the imagination
mode while doing this, the result might well be just what you described,
uttered out loud: "35 millimeters --- no, wait, 30, 32 millimeters." Each
number that is uttered results in a sense of length, which is judged against
the sense of length derived from looking at or feeling the object, and if
there is a difference it is corrected.

Beware, by the way, of saying that the visual or kinesthetic sense of length
is necessarily a reference signal. It is a perception (although it could,
through being remembered, become a reference signal). The situation is
analogous to saying that the dot is the reference position that the position
of the knot is being compared with, in the rubber-band demo. You can show
that this is wrong by asking the person to keep the knot one inch left of
the dot, and seeing that this also can easily be done. I could easily choose
to describe a length that was twice as large as the perceived length of the
object. Then I would be controlling for a verbally-evoked sense of length
that was twice as large as the object-evoked sense of length. Of course I
could also control for the two senses of length being the same, and mentally
multiply the resulting number by two, like a fisherman reporting the size of
the fish he caught. The poijnt is that the reference condition is not
necessarily "verbally-evoked sense of length EQUALS perceived sense of length."

Note that the "talking ruler" mode of behavior does NOT ANYWHERE involve a
direct translation from the perceived object into a verbal description of
size. It involves a closed loop in which the verbal description is varied
until the sense of size it evokes is the same as the sense of size that the
object evokes in perception. This control loop looks, from the outside, like
a stimulus-response pair, but that is just the old illusion.

路路路

------------------------------
My impression is that a die-hard behaviorist might start out by assuming
that unless proven otherwise, all overt behavior consists of responses to
stimuli. If this behaviorist started to see some merit in PCT, the next
stage would be to admit that there might be special cases in which there is
a closed loop control system working, although there would still be a
majority of cases in which the basic response-to-stimuli arrangement
existed. After enough experience with control situations, this behaviorist
might gradually come to see that a rather large number of situations really
do involve closed-loop control, but that there are still enough obvious
examples of input-output behaviors that we can't afford to ignore them.

When this behaviorist finally comes to the position I hold, he will conclude
that unless proven otherwise, there is simply no behavior that is not part
of a closed-loop control process, and that organisms never "respond to
stimuli." NEVER.

Best,

Bill P.

[From Rick Marken (970228.2230 PST)]

Bill Powers (970228.1826 MST) --

Thanks. That's the best birthday present I've ever gotten.

Love

Rick

[From Bruce Abbott (970302.1515 EST)]

Bill Powers (970228.1826 MST) --

I get the impression from reading this post that we are not on quite the
same wavelength. I am _not_ asserting that _any_ observable _behavior_
occurs open-loop, but only that certain components of the system either are
themselves open-loop systems, or can be modeled as such for certain analytic
purposes. To the extent that this is true, conventional methods can yeild
useful and accurate information about them. My argument has _not_ been that
some behaviors are emitted open-loop, while others are products of control
system operation. Rather, my reasoning was developed specifically to
addresses the question of whether an open-loop analysis of some components
can yield useful information about the system's characteristics (e.g.,
input-output PIF under normal operating conditions). I concluded that it could.

You raise an important point, which is that even system components that
appear to be operating open loop by external appearances are quite likely
themselves to be control systems, or at least closed-loop systems. This
seems to me quite natural because of the relative stability of such systems,
not to mention the potential speed and bandwidth advantages. I have no
problem envisioning how many associated systems, such as those involved in
memory access, may rely on feedback mechanisms, perhaps (in the case of
memory access) operating something like a phase-locking circuit in which a
set of "probe" signals act as a sort of reference vector.

Beware, by the way, of saying that the visual or kinesthetic sense of length
is necessarily a reference signal. It is a perception (although it could,
through being remembered, become a reference signal).

Yes, I was keenly aware of this when thinking about the loop I diagramed
that accomplishes the parameter tuning for the "translator function." Both
the perception of line length and the experimenter's verbal statements
giving the measured values of line length are perceptual inputs to the
learner, as is the learner's awareness of her own judgements. You will note
that the line length enters my diagram as a sensory input, not as a reference.

My impression is that a die-hard behaviorist might start out by assuming
that unless proven otherwise, all overt behavior consists of responses to
stimuli. If this behaviorist started to see some merit in PCT, the next
stage would be to admit that there might be special cases in which there is
a closed loop control system working, although there would still be a
majority of cases in which the basic response-to-stimuli arrangement
existed. After enough experience with control situations, this behaviorist
might gradually come to see that a rather large number of situations really
do involve closed-loop control, but that there are still enough obvious
examples of input-output behaviors that we can't afford to ignore them.

When this behaviorist finally comes to the position I hold, he will conclude
that unless proven otherwise, there is simply no behavior that is not part
of a closed-loop control process, and that organisms never "respond to
stimuli." NEVER.

Well, the first organisms evolved out of the muck and after several billion
years of continued evolution we had invertebrates, then the vertebrates,
then large-brained vertebrates. A few million years later some of these
evolved into behaviorists, and now a few of them are beginning to take those
first steps toward evolving into control theorists. Evolution takes a long
time -- one has to be patient!

Regards,

Bruce

[From Bruce Gregory (970302.2015 EST)]

Bruce Abbott (970302.1515 EST)

>Bill Powers (970228.1826 MST) --

>When this behaviorist finally comes to the position I hold, he will conclude
>that unless proven otherwise, there is simply no behavior that is not part
>of a closed-loop control process, and that organisms never "respond to
>stimuli." NEVER.

Well, the first organisms evolved out of the muck and after several billion
years of continued evolution we had invertebrates, then the vertebrates,
then large-brained vertebrates. A few million years later some of these
evolved into behaviorists, and now a few of them are beginning to take those
first steps toward evolving into control theorists. Evolution takes a long
time -- one has to be patient!

One way to try to speed the evolutionary process is to adopt the
view that there is no behavior that is not part of a closed-loop process,
and each time you think you have uncovered an example of
open-loop behavior, tell yourself that you haven't worked hard
enough (or been clever enough) to identify the loop involved. At least,
that's the approach I use.

Bruce Gregory

[From Rick Marken (960906.0800)]

Bruce Abbott (960906.0855 EST)--

Here is Nobel laureate Herbert Simon's version [of a profound observation
about behavioral simplicity]:

A man, viewed as a behaving system, is quite simple. The apparent
complexity of his behavior over time is largely a reflection of the
complexity of the environment in which he finds himself.
[Simon, H. A. (1969). _The sciences of the artificial_. Cambridge, MA:

          MIT Press, p. 25]

So you see, Bruce (and Rick, maybe), you're in good company.

Although he's a fellow member of the International Jewish Conspiracy (I'm
still waiting for my check, guys), I don't count Herb as "good company".
What he says above sounds good but I think you'll find that it's based on a
version of the old input-output view of behavior: complex inputs lead to
complex outputs while the law relating input to output is simple.

The Nobel committee has made many perculiar selections, especially when
awarding the prize to social scientists like Simon. The Nobel prize will
regain its luster for me when it is awarded for the only discovery in the
social sciences that merits such recognition: the discovery of the fact that
behavior is the control of perception. I'll probably get my check from the
IJC before that happens;-)

RSM

[From Bill Powers (960906.1500 MDT)]

Francisco arocha, 96/09/06, 15.27, EST]
Bruce Abbott (960906.1255 EST) --

Interesting discussion of Herbert Simon. If you read his description, he's
talking about feedback control, but if you read his explanation, it's
stimulus-response. The critical part comes when he says that this is a
goal-directed process -- and then explains the goal as being a motor program
produced out of perceptions of the environment.

What is the intellectual barrier that prevents so many scientists from
taking that last step?

Best,

Bill P.

路路路

------------------------------------------------------------------------

[From Dag Forssell (960907.0900)]

[Bill Powers (960906.1500 MDT)]

What is the intellectual barrier that prevents so many scientists >from

taking that last step?

I think it is called perceptual control; once an idea has been woven into a
person's personal web of understanding it controls for its own continued
existence, resisting disturbances. See Rick Marken's paper: The
Hierarchical Behavior of Perception.

Simon's dichotomy is by no means unique or even unusual. In fact it is
commonplace. Most clinical psychologists and lay people alike clearly
recognize that people do what they want to do and resist disturbances in the
process. Thus most people on the planet anticipate William T. Powers, and
PCT is no great deal. Let's accentuate the positive! But the same people
invariably explain in terms of stimulus-response and are thus clueless as to
the true nature of behavior. Moreover, they will defend their clueless
stimulus-response thinking to the death. But let us not dwell on the negative.

Last month, while in Sweden, I read daily in the newspapers of difficulties
people suffer and how they cannot be expected to do anything about it
themselves. Society must do things for them. Pervasive stimulus-response
thinking leads to a victim mentality that I find nauseating.

Christine and I also visited with an old friend, a lady psychologist who
works in the trenches with school children and their parents and gets
involved in all kinds of developmental and social problems.

I remember from past visits how this friend defended the science of
psychology, noting "that's why we have the scientific method". I refrained
from coming back with the idea that the scientific method is applied in an
inappropriate way when psychologists study living organisms.

This time I learned from her that you must always accept whatever people say
and think, because that is their truth, regardless of how screwy it seems to
you, and that the only time you can accomplish anything is when people are
concerned and unhappy with where they are and what's going on in their
lives. Sounded like Ed Ford reincarnate.

Rick has suggested before on this net that it does not matter what theories
psychologists learn in school (and there is a great variety of those); when
they get into the trenches, they reorganize and do what works -- still
paying homage to the theories.

Let us always remember when we bash psychologists on CSGnet, that it is the
"scientific" theorizing we bash without mercy, not wise people in the
trenches -- people just as wise as wise people have always been without
valid theories. The clinical practice of psychology today has nothing
whatsoever to do with the theorizing of psychology.

The practice of psychology of tomorrow will be well supported by perceptual
control theory and psychologists and lay people alike will find the subject
matter much simplified and will be much wiser and more clear than they can
be today, based on experience alone.

Bruce Abbott and many before him and many to come will continue to point to
people who sense that people are indeed goal driven etc. But it is a
mistake in this forum to suggest that we be impressed by such simple,
elementary, commonplace observation and consider it to be good enough to
build on. Unless there is an explicit, correct explanation of perceptual
feedback control, we do not have anything to build on, quite the contrary --
what we have is yet another authority who thinks he or she knows something.

In this forum, for the purpose of spreading a correct understanding of what
behavior is and how it works, we are better off building on people who are
new to the field and do not claim prior understanding, but who are curious
for some reason -- dissatisfaction with something. Bruce came to study PCT
because of dissatisfaction. By the way, Bruce, you post like a PCTer most
of the time nowadays. (This last post seems a throwback.)

For this reason, I think PCT must be promoted outside of the ranks of the
"social scientific" establishment. Once the public catches on, the
"scientists" will die off in oblivion.

Bill Powers: in Flagstaff, I lent you _Information Anxiety_ by Richard Saul
Wurman. Have you looked at it? What did you think of his observations of
what it takes for people to pay attention and understand new information?
There are no claims of science here, but I thought he made good sense.

Best, Dag