Perception and Input

CSGnet Archive CSG2001
1

[From Rick Marken (2001.09.07.1200)]

Bruce Gregory (2001.0907.1335)]

I would be happier if we talked in terms of inputs and outputs, which
are well-defined in PCT, and we avoided the term "perception," which
as far as I can tell is neither well-defined, nor necessary for the model.

I don't see how it can be avoided. The model says that external input
(in the energy at the sensory receptors) is transformed, via perceptual
functions, into perceptual signals. A perceptual signal is very well
defined: it's the time varying, scalar output of a perceptual function.
Inputs to a perceptual function are either physical variables or (for
higher order perceptions) perceptual signals that are the outputs of
lower level perceptual functions.

Some inputs are associated with what most people would call perceptions,
but others are not.

The input variable, in PCT, is energy at the sensory surface. A control
system can't do anything with this input until it exists in a form that
can be controlled: that is, in a form that can be compared to a neural
reference signal. That form, according to PCT, is a neural perceptual
signal. The perceptual signal is, therefore, the output of a function
that transforms a physical input variable (or variables) into a
perceptual signal.

I would have no objection to the statement, "It's all input control."

But we don't control input. We control _functions_ of input. The actual
input at the eye of a fielder catching a ball is an array of
electromagnetic energy varying over the relevant space (the area of the
retina) and time in terms of intensity, wavelength and phase. It's not
this changing, multidimensional input array that is controlled; it's
functions of that array (perceptions) that are controlled. According to
PCT (with excellent support from research and modeling), it's all
_perceptual_ control, where the perceptions controlled are a _function
of_ input.

Best regards

Rick

···

---
Richard S. Marken, Ph.D.
MindReadings.com
10459 Holman Ave
Los Angeles, CA 90024
Tel: 310-474-0313
E-mail: marken@mindreadings.com

2

[From Bruce Gregory (2001.0907.1600)]

[From Rick Marken (2001.09.07.1200)]

You raise several points that I would like to address

I don't see how it can be avoided. The model says that external input
(in the energy at the sensory receptors) is transformed, via perceptual
functions, into perceptual signals. A perceptual signal is very well
defined: it's the time varying, scalar output of a perceptual function.
Inputs to a perceptual function are either physical variables or (for
higher order perceptions) perceptual signals that are the outputs of
lower level perceptual functions.

I would simply say that the input to a control system is very well defined.
It isn't necessary to call it a perceptual signal, but of course you can
call it anything you like.

> Some inputs are associated with what most people would call perceptions,
> but others are not.

The input variable, in PCT, is energy at the sensory surface. A control
system can't do anything with this input until it exists in a form that
can be controlled: that is, in a form that can be compared to a neural
reference signal. That form, according to PCT, is a neural perceptual
signal. The perceptual signal is, therefore, the output of a function
that transforms a physical input variable (or variables) into a
perceptual signal.

At the lowest level (intensity?) this is true. As far as I know, it is not
true at higher levels because all inputs to a higher level system are
already neural signals.

> I would have no objection to the statement, "It's all input control."

But we don't control input. We control _functions_ of input.

You are using the word input in a different way than I am. The input to a
control system is a neural signal and the control system is controlling
that signal, no? In a hierarchical control system the vast preponderance of
inputs are neural signals. Only the lowest level inputs presumably involve
transducers. These neural signals can be referred to as functions of
exogenous inputs.

The actual
input at the eye of a fielder catching a ball is an array of
electromagnetic energy varying over the relevant space (the area of the
retina) and time in terms of intensity, wavelength and phase. It's not
this changing, multidimensional input array that is controlled; it's
functions of that array (perceptions) that are controlled.

Yes indeed. But most of the processing does _not_ involve transducers. It
is _these_ higher-level inputs that are controlled.

According to
PCT (with excellent support from research and modeling), it's all
_perceptual_ control, where the perceptions controlled are a _function
of_ input.

We are not disagreeing about the mechanism only about the labels, at least
as far as I can see. In PCT certain inputs to control circuits are called
perceptions. These inputs are controlled by the circuits. Some inputs are
functions of "external" forces. Most inputs are not. If no one is confused
by calling these inputs perceptions, fine. I believe that I understand how
PCT using the term perception. It is not clear to me however, what role, if
any, perceptions that are _not_ controlled play in PCT. In other words, how
do you model an uncontrolled perception? And why would you bother?

3

[From Rick Marken (2001.09.07.1400)]

Bruce Gregory (2001.0907.1600)

You are using the word input in a different way than I am. The input to a
control system is a neural signal and the control system is controlling
that signal, no?

I think you would see the problem if you drew a diagram of a control system.

       I11 I21

    qi--->|f1()|--->|f2()|
     ^ | |

···

     > I12 | I22

     > v v
     > >C1| |C2|
     > > >
     > > >
     ------------------

Here are two control systems. One (made up of perceptual function f1()
and comparator C1) gets what I call "input" from the environment (line
I11). The other (made up of perceptual function f2() and comparator C2)
gets "input" from the first control system (line I21). I think you want
to call the lines going into the comparators of the two control systems
(lines I12 and I22) "inputs" or "input variables". I12 and I22 are what
we (PCTers) call "perceptual variables". We do this to distinguish these
variables from what we call "inputs", which are the inputs to the
functions (f1() and f2()) that produce these variables (I11 and I21). If
you want to call I12 and I22 "inputs", then what do you call variables
I11 and I21? I suggest that the best idea is to call I12 and I22
"perceptual variables" because they are functions of other variables
(I11 and I21 respectively). I also suggest that I11 and I21 be called
"input variables" because they are inputs in the sense that they _go
into_ the control system via the perceptual function. In other words, I
suggest that we stick with existing PCT terminology, which is nice and
clear and consistent with conventional understanding.

Best regards

Rick

--
Richard S. Marken, Ph.D.
MindReadings.com
10459 Holman Ave
Los Angeles, CA 90024
Tel: 310-474-0313
E-mail: marken@mindreadings.com

4

[from Jeff Vancouver (2001.09.07.1700)]

[From Bruce Gregory (2001.0907.1600)]
In
other words, how
do you model an uncontrolled perception? And why would you bother?

Not that I don't think you have made some good points, but what is an
uncontrolled perception? If I modeled a system with four sensory inputs,
which are combined via second level input (or perceptual) functions into
some finite set of second level perceptions (where the number of perceptions
is determined by the number of input functions), I am at the same time not
defining an infinite number of other perceptions that could have been
created via some infinite number of input functions (assuming the sensory
inputs are continuous, of course).

Could some first level perception arise that is never used? That seems an
unlikely waste of resources. Could one become obsolete via evolution? More
likely. Could only some incorporate the resulting signals into control
systems? ESP maybe? Well that could be picked up in the test (although a
non-endowed researcher may have a tough time hypothesizing the variable, or
get hung up and the sensor finding step).

It is getting late.

Later,

Jeff

5

[From Bruce Gregory (2001.0907.1722)]

Rick Marken (2001.09.07.1400)

I suggest that the best idea is to call I12 and I22
"perceptual variables" because they are functions of other variables
(I11 and I21 respectively). I also suggest that I11 and I21 be called
"input variables" because they are inputs in the sense that they _go
into_ the control system via the perceptual function. In other words, I
suggest that we stick with existing PCT terminology, which is nice and
clear and consistent with conventional understanding.

I would never suggest that we change existing PCT terminology. I agree that
it is clear. I have my doubts that it is consistent with conventional
understanding (because of the reasons I outlined). But the situation will
no doubt change when PCT becomes the conventional model.

6

[From Bruce Gregory (2001.0907.1736)]

Jeff Vancouver (2001.09.07.1700)

> [From Bruce Gregory (2001.0907.1600)]
> In
> other words, how
> do you model an uncontrolled perception? And why would you bother?

Not that I don't think you have made some good points, but what is an
uncontrolled perception?

If I hesitated a guess, it would be that most of our perceptions are
uncontrolled. As I look out my office window a perceive trees, laws, people
playing tennis, parked cars,... I am not controlling any of these perceptions.

If I modeled a system with four sensory inputs,
which are combined via second level input (or perceptual) functions into
some finite set of second level perceptions (where the number of perceptions
is determined by the number of input functions), I am at the same time not
defining an infinite number of other perceptions that could have been
created via some infinite number of input functions (assuming the sensory
inputs are continuous, of course).

Yes. And....?

Could some first level perception arise that is never used?

Again, most are not used.

That seems an
unlikely waste of resources.

Not to me. I have to exercise considerable discretion over the perceptions
I control. But that doesn't mean I could eliminate most of them without
severely crippling my chances for survival. I am not controlling my
perception of the distance between me and the cars outside my window parked
below me. When I get on the highway, however, I definitely do control the
distance between my car and other cars.

Could one become obsolete via evolution? More
likely. Could only some incorporate the resulting signals into control
systems? ESP maybe? Well that could be picked up in the test (although a
non-endowed researcher may have a tough time hypothesizing the variable, or
get hung up and the sensor finding step).

I take it that you do not distinguish between perceptions that you might
control and perceptions that you are in fact controlling. I think this
might be a mistake. I admit that the model is much simpler if you simply
ignore uncontrolled perceptions. Indeed that seems to be true of most PCT
models. That does not mean, however, that such perceptions do not exist.
For example, I don't control my perception of the weather, but I allow that
perception to play a role in deciding which perceptions I do control. I
assume you behave in a similar way.

7

[from Jeff Vancouver (2001.09.09.14:30)]

[From Bruce Gregory (2001.0907.1736)]

> Jeff Vancouver (2001.09.07.1700)
>
>
> > [From Bruce Gregory (2001.0907.1600)]
> > In
> > other words, how
> > do you model an uncontrolled perception? And why would you bother?
>
>Not that I don't think you have made some good points, but what is an
>uncontrolled perception?

If I hesitated a guess, it would be that most of our perceptions are
uncontrolled. As I look out my office window a perceive
trees, laws, people
playing tennis, parked cars,... I am not controlling any of
these perceptions.

You seem to think that a perception not involved in a system (ECU) with
error is not being controlled. If those trees, laws, people playing tennis
or parked cars were hurdling in your direction, you would likely duck (or in
the case of laws, figure out a way around them or to change them)?

>Could some first level perception arise that is never used?

Again, most are not used.

None of the examples you mentioned are anywhere near the first level.

>That seems an
>unlikely waste of resources.

Not to me. I have to exercise considerable discretion over
the perceptions
I control.

This is a different issue altogether. Given our limited degrees of freedom,
we cannot always act on the errors we have. Determining how we allocate
those resources is a critical issue as far as I am concerned.

I take it that you do not distinguish between perceptions
that you might
control and perceptions that you are in fact controlling. I think this
might be a mistake.

Again, is "not controlling" no error or error on standby until resources are
freed up, or is it more (or less)?

I have a colleague who is interested in the "units of perception" and how
they change given context. For instance, the question might be whether you
perceived that tree always, or only when you are approaching it (and thus
want to avoid hitting it) or contemplating nature, etc. (i.e., controlling
something for which the tree is an aspect).

I admit that the model is much simpler if
you simply
ignore uncontrolled perceptions. Indeed that seems to be true
of most PCT
models. That does not mean, however, that such perceptions do
not exist.
For example, I don't control my perception of the weather,
but I allow that
perception to play a role in deciding which perceptions I do
control. I
assume you behave in a similar way.

If you are using your perception of the weather to help you control other
perceptions, is not the weather part of those other perceptions? In other
words, I am controlling that I do not get wet. When I perceive that it is
raining outside, I do things (like not go out, get an unbrella) to maintain
my not wet perception. My perception of the weather substantially affects
what I do, hence it is used by my control systems. Hence, it is an aspect
of perceptions I control. I think the interesting question is what is the
nature of the input functions involved (i.e, what does "aspect" mean in my
previous sentence) in this type of scenario. Makes for a great modeling
problem.

Bruce, despite what I say, I think you make your point wonderfully. Many
psychologists are going to find PCT's understanding of perception to be
possibly very odd.

Later,

Jeff

8

[From Bruce Gregory (2001.0909.1542)]

Jeff Vancouver (2001.09.09.14:30)

You seem to think that a perception not involved in a system (ECU) with
error is not being controlled.

Yes, that's exactly what I think.

If those trees, laws, people playing tennis
or parked cars were hurdling in your direction, you would likely duck (or

in

the case of laws, figure out a way around them or to change them)?

Yes, but that does not mean I am controlling my perception of them right
now. At least the Test would not reveal that I am controlling for their
present location. Unless, I totally fail to understand the Test, which is
quite possible.

This is a different issue altogether. Given our limited degrees of

freedom,

we cannot always act on the errors we have. Determining how we allocate
those resources is a critical issue as far as I am concerned.

You seem to be saying that we are controlling all of our perceptions. This
appears to make the notion of a controlled perception of little utility.
Perhaps I missed your point.

Again, is "not controlling" no error or error on standby until resources

are

freed up, or is it more (or less)?

So you are saying that we are controlling all our perceptions at all times,
but are limited in our effectiveness by our resources. This sounds a bit
like theology to me. How might I test your conjecture that all perceptions
are controlled, but some are controlled better than others? I presume that
if the Test fails in some situation, it simply means that I am controlling
that particulat at very low gain.

If you are using your perception of the weather to help you control other
perceptions, is not the weather part of those other perceptions? In other
words, I am controlling that I do not get wet. When I perceive that it is
raining outside, I do things (like not go out, get an unbrella) to

maintain

my not wet perception. My perception of the weather substantially affects
what I do, hence it is used by my control systems. Hence, it is an aspect
of perceptions I control.

So it's not only all perception, it's all controll too.

Bruce, despite what I say, I think you make your point wonderfully. Many
psychologists are going to find PCT's understanding of perception to be
possibly very odd.

Psychologists won't be alone in their puzzlement.

9

i.kurtzer (2001.09.09.2200)

[From Bruce Gregory (2001.0909.1542)]

Jeff Vancouver (2001.09.09.14:30)
>
>
> You seem to think that a perception not involved in a system (ECU) with
> error is not being controlled.

Yes, that's exactly what I think.

> If those trees, laws, people playing tennis
> or parked cars were hurdling in your direction, you would likely duck

(or

in
> the case of laws, figure out a way around them or to change them)?

Yes, but that does not mean I am controlling my perception of them right
now. At least the Test would not reveal that I am controlling for their
present location. Unless, I totally fail to understand the Test, which is
quite possible.

The TEST does not infer reference signals by noting actions.
Rather the Test requires the application of a independent disturbance to a
proposed controlled variable and notes the stability of that variable.
One could apply a perturbation to that variable until the output dropped to
zero. This would not necessarily mean that control was not in place, i.e.
"right now", but could reflect that the input now matches the reference; of
course, this could be a very broad reference.

i.

10

[From Bruce Gregory (2001.0910.0715)]

i.kurtzer (2001.09.09.2200)

The TEST does not infer reference signals by noting actions.
Rather the Test requires the application of a independent disturbance to a
proposed controlled variable and notes the stability of that variable.
One could apply a perturbation to that variable until the output dropped

to

zero. This would not necessarily mean that control was not in place, i.e.
"right now", but could reflect that the input now matches the reference;

of

course, this could be a very broad reference.

O.K. I apparently have a very broad reference for the location of all the
vehicles in the world. At least while I am working in my office. In that
case, it appears that I am indeed controlling everything. No wonder PCT
seems odd to most people.

11

i.kurtzer (2001.09.10.1100)

[From Bruce Gregory (2001.0910.0715)]

i.kurtzer (2001.09.09.2200)
>
> The TEST does not infer reference signals by noting actions.
> Rather the Test requires the application of a independent disturbance to

a

> proposed controlled variable and notes the stability of that variable.
> One could apply a perturbation to that variable until the output dropped
to
> zero. This would not necessarily mean that control was not in place,

i.e.

> "right now", but could reflect that the input now matches the

reference;

of
> course, this could be a very broad reference.
>
O.K. I apparently have a very broad reference for the location of all the
vehicles in the world. At least while I am working in my office. In that
case, it appears that I am indeed controlling everything. No wonder PCT
seems odd to most people.

PCT doesn't seem odd to most people, that have even heard of it, because of
the issue of potentially controlling so many things.
Rather, there is very little developed research. And what research there is
is reliably misunderstood, with complete meaning reversals, since control is
ruled out
beforehand in both the concepts and experimental designs. I suggest if you
find this disagreeable you stop arguing about
whether "perceptual signal" means "perception" to everyone and just do some
research, any research. 20 years of that we may be in a better spot.

i.

12

[from Jeff Vancouver (2001.09.10.10:20)]

Here is another issue. Suppose we observe someone being hit by a car as
they cross a street. This is recorded in episodic memory. Why? It may

be

useful for control later, but are we creating a recording of all we see

just

in case? How does that correspond with PCT? This is really out of my

area.

I vaguely remember talking about these issues in my cognitive psychology
class, and I am not sure what is suspected, supported, or even not

supposed.

In my armchair musing on this I have thought about the reliable difference
in recalled vs. recognized items in a memory task.
The number of recognized items is always larger than the items one can
recall. Maybe the recalled items reflect a higher level about that
genreated the reference while the recognized items do not require a higher
level output but only a fixed stored reference. There also the "levels of
processing" stories that could be revisited.
I suggest that this is EXCATLY what grad students and senior honor theses
are for.

i.

13

[from Jeff Vancouver (2001.09.10.10:20)]

[From Bruce Gregory (2001.0909.1542)]

Jeff Vancouver (2001.09.09.14:30)
>
>
> You seem to think that a perception not involved in a
system (ECU) with
> error is not being controlled.

Yes, that's exactly what I think.

So you are saying that when the furnace is off (because there is no error),
the perception of the room temperature is not being controlled?

> If those trees, laws, people playing tennis
> or parked cars were hurdling in your direction, you would
likely duck (or
in
> the case of laws, figure out a way around them or to change them)?

Yes, but that does not mean I am controlling my perception of
them right
now. At least the Test would not reveal that I am controlling
for their
present location. Unless, I totally fail to understand the
Test, which is
quite possible.

The Test would be very difficult to apply because it requires creating
disturbances at a couple of places. There are many potentially controlled
variables that we would not be able to manipulate (i.e., disturb). If there
are no disturbances, there is no error (unless there was always error or the
reference signal changed).

> This is a different issue altogether. Given our limited degrees of
freedom,
> we cannot always act on the errors we have. Determining
how we allocate
> those resources is a critical issue as far as I am concerned.

You seem to be saying that we are controlling all of our
perceptions. This
appears to make the notion of a controlled perception of
little utility.
Perhaps I missed your point.

I am uncomfortable with this thread because of the following: Suppose you
are playing the coin game. You have a pattern in mind that another is
attempting to determine. It is not that the coins form a square. Now
suppose the other, asks, in an attempt to clarify understanding of the game,
"so let me get this straight, these coins form a square and you have no
problem with that, but you are not controlling for the coins to form a
square." You answer yes. To get to that yes, you likely confirm that the
coins form a square. That perception of the shape of the coins, which might
have nothing to do with the pattern you are controlling in the game, is used
in the conversation because it relates to an other perception that the
person understand the game. Or even that in the course of the game, you
smile to yourself because you suspect the other player is testing for a
square shape. In these cases, the square perception does not appear to feed
into a comparator, but only into higher level input functions that create
perceptions of qualitatively different content. So the question is, does a
perception have to feed into a comparator to be a perception in PCT terms?
Or is it really tied to a comparator with a gain of 0 (you could have made a
square shape the pattern for the game)? I do not know the answer to these
questions.

> Again, is "not controlling" no error or error on standby
until resources
are
> freed up, or is it more (or less)?

So you are saying that we are controlling all our perceptions
at all times,
but are limited in our effectiveness by our resources. This
sounds a bit
like theology to me. How might I test your conjecture that
all perceptions
are controlled, but some are controlled better than others? I
presume that
if the Test fails in some situation, it simply means that I
am controlling
that particulat at very low gain.
>

I think this issue does create a problem for the usefulness of the theory,
which is why the systems that control the allocation of resources are
critical to understand.

>
> If you are using your perception of the weather to help you
control other
> perceptions, is not the weather part of those other
perceptions? In other
> words, I am controlling that I do not get wet. When I
perceive that it is
> raining outside, I do things (like not go out, get an unbrella) to
maintain
> my not wet perception. My perception of the weather
substantially affects
> what I do, hence it is used by my control systems. Hence,
it is an aspect
> of perceptions I control.

So it's not only all perception, it's all controll too.

I understand your issue (I think). See above.

> Bruce, despite what I say, I think you make your point
wonderfully. Many
> psychologists are going to find PCT's understanding of
perception to be
> possibly very odd.

Psychologists won't be alone in their puzzlement.

Here is another issue. Suppose we observe someone being hit by a car as
they cross a street. This is recorded in episodic memory. Why? It may be
useful for control later, but are we creating a recording of all we see just
in case? How does that correspond with PCT? This is really out of my area.
I vaguely remember talking about these issues in my cognitive psychology
class, and I am not sure what is suspected, supported, or even not supposed.

Later,

Jeff

14

[From Rick Marken (2001.09.10.0830)]

Jeff Vancouver (2001.09.10.10:20)

does a perception have to feed into a comparator to be a perception
in PCT terms?

No. A perception, in PCT, is defined as the existence of a neural
current in the output of a perceptual function. Perceptions are
_variables_. Many of these variables are controlled but many are not, at
any particular time anyway. A perceptual variable that is _not_
controlled may be the perceptual input to a one-way control system whose
reference signal is set to zero (and the value of the perceptual
variable is less than zero) so that variations in the perception "don't
matter" to the control system; or it may be a perceptual signal that
doesn't enter a control system. But I think the latter would is
unlikely: Why would the nervous system develop the ability to perceive a
variable that it would never have to control?

I think a perception of, say, the sunset, is an uncontrolled perception.
But a sunset is a collection of perceptual _variables_ -- brightnesses,
distances, colors, relationships, configurations -- that _can_ be
controlled in other circumstances, for example, when paining a sunset.

I think this whole discussion of whether or not there are controlled and
uncontrolled perceptions would benefit from remembering that
perceptions, in PCT, are _variables_. We don't control "cars", for
example, but variable perceptual aspects of cars: their speed, color,
shape, location, direction, etc. All of these are variables that might
also be controlled when they are _not_ aspects of a car. For example,
you continue to control speed when you get out of the car and walk. When
seen from this perspective I think you can see that it is probably quite
unlikely that there are any perceptions (perceptual variables) that are
not controlled (or, at least, controllable).

Best regards

Rick

···

--
Richard S. Marken, Ph.D.
MindReadings.com
10459 Holman Ave
Los Angeles, CA 90024
Tel: 310-474-0313
E-mail: marken@mindreadings.com

15

[from Jeff Vancouver (2001.09.10.11:50)]

It is about time you entered this thread. See you are needed!

Jeff

···

Jeff Vancouver (2001.09.10.10:20)

> does a perception have to feed into a comparator to be a perception
> in PCT terms?

No. A perception, in PCT, is defined as the existence of a neural
current in the output of a perceptual function. Perceptions are
_variables_. Many of these variables are controlled but many
are not, at
any particular time anyway. A perceptual variable that is _not_
controlled may be the perceptual input to a one-way control
system whose
reference signal is set to zero (and the value of the perceptual
variable is less than zero) so that variations in the
perception "don't
matter" to the control system; or it may be a perceptual signal that
doesn't enter a control system. But I think the latter would is
unlikely: Why would the nervous system develop the ability to
perceive a
variable that it would never have to control?

I think a perception of, say, the sunset, is an uncontrolled
perception.
But a sunset is a collection of perceptual _variables_ --
brightnesses,
distances, colors, relationships, configurations -- that _can_ be
controlled in other circumstances, for example, when paining a sunset.

I think this whole discussion of whether or not there are
controlled and
uncontrolled perceptions would benefit from remembering that
perceptions, in PCT, are _variables_. We don't control "cars", for
example, but variable perceptual aspects of cars: their speed, color,
shape, location, direction, etc. All of these are variables that might
also be controlled when they are _not_ aspects of a car. For example,
you continue to control speed when you get out of the car and
walk. When
seen from this perspective I think you can see that it is
probably quite
unlikely that there are any perceptions (perceptual
variables) that are
not controlled (or, at least, controllable).

Best regards

Rick
--
Richard S. Marken, Ph.D.
MindReadings.com
10459 Holman Ave
Los Angeles, CA 90024
Tel: 310-474-0313
E-mail: marken@mindreadings.com

16

[From Bruce Gregory (2001.0910.1209)]

i.kurtzer (2001.09.10.1100)

PCT doesn't seem odd to most people, that have even heard of it, because of
the issue of potentially controlling so many things.
Rather, there is very little developed research. And what research there is
is reliably misunderstood, with complete meaning reversals, since control is
ruled out
beforehand in both the concepts and experimental designs. I suggest if you
find this disagreeable you stop arguing about
whether "perceptual signal" means "perception" to everyone and just do some
research, any research. 20 years of that we may be in a better spot.

Thanks. I'll get back to you in twenty years.

17

[From Bruce Gregory (2001.0910.1217)]

Rick Marken (2001.09.10.0830)

I think a perception of, say, the sunset, is an uncontrolled perception.
But a sunset is a collection of perceptual _variables_ -- brightnesses,
distances, colors, relationships, configurations -- that _can_ be
controlled in other circumstances, for example, when paining a sunset.

So perceptions are not controlled, only perceptual variables are
controlled. I do not control my perception of Mt. Everest, I may control my
perception of the speed at which Mt. Everest is approaching me.

18

[From Bruce Gregory (2001.0910.13110]

[From Rick Marken (2001.09.10.0830)]

I think this whole discussion of whether or not there are controlled and
uncontrolled perceptions would benefit from remembering that
perceptions, in PCT, are _variables_. We don't control "cars", for
example, but variable perceptual aspects of cars: their speed, color,
shape, location, direction, etc. All of these are variables that might
also be controlled when they are _not_ aspects of a car. For example,
you continue to control speed when you get out of the car and walk. When
seen from this perspective I think you can see that it is probably quite
unlikely that there are any perceptions (perceptual variables) that are
not controlled (or, at least, controllable).

This helps a great deal. If I read PCT as Perceptual-variable Control
Theory, the model is much clearer. I would prefer not to blur the
distinction between perceptions and perceptual variables. They are clearly
two quite different constructs.

19

[From Rick Marken (2001.09.10.1111)]

Bruce Gregory (2001.0910.13110) --

If I read PCT as Perceptual-variable Control Theory, the model is
much clearer.

Yes. Calling PCT perceptual variable control theory might help make
clear what PCT is about. If perceptions were not variable there would be
no need for (or possibility of) controlling them; they would always be
in the same state, no matter what.

I would prefer not to blur the distinction between perceptions and
perceptual variables. They are clearly two quite different constructs.

I think it is good not to blur the distinction. I think this can be done
by being clear about what the distinction is. I think the distinction is
this: the word "perception" refers to a particular state of one or more
perceptual variables. So when you point to a "car" (or "Mt. Everest") as
a "perception" you are talking about a particular state of a set of
perceptual variables: shape (Lexus sedan), color (black), speed (zero),
location (neighbor's garage), orientation (trunk towards viewer),
condition (clean), etc.

Best regards

Rick

···

--
Richard S. Marken, Ph.D.
MindReadings.com
10459 Holman Ave
Los Angeles, CA 90024
Tel: 310-474-0313
E-mail: marken@mindreadings.com

20

from one of little experience and possibly scant insight on PCT I thought
PCT may have some real possibilities in terms of landscape perception and
of building perception and could be applied to bring this area of perception
in my field forward. I know this is not a comment into the mechanics of PCT
but rather an applied possibility, and hope it still adds to the discussion.

The area of landscape perception and preference is quite interesting in that
it is often conceptualised as static, with little interaction between person
and environment.

Initally People could be tested to see if they are controlling some physical
form such as organistion or complexity in the scene in terms of there
evaluation of it.

But also I think people have higher reference values for building and
landscapes possible related to principle level reference values such as does
the landscape facilitate 'interest', or ' relaxation', or 'importance', or
'wonder' and depending on what a person is controlling when the _percieve_ a
building or landscape the preferences/ evaluations will vary.

In this way, and many others, I feel like PCT has a lot to offer my field in
terms of perception. It is the task however of getting out there and doing
it....hopefully I'll get there

Rohan Lulham.

···

-----Original Message-----
From: Control Systems Group Network (CSGnet)
[mailto:CSGNET@LISTSERV.UIUC.EDU]On Behalf Of Bruce Gregory
Sent: Tuesday, September 11, 2001 2:17 AM
To: CSGNET@LISTSERV.UIUC.EDU
Subject: Re: Perception and Input

[From Bruce Gregory (2001.0910.1217)]

Rick Marken (2001.09.10.0830)

I think a perception of, say, the sunset, is an uncontrolled perception.
But a sunset is a collection of perceptual _variables_ -- brightnesses,
distances, colors, relationships, configurations -- that _can_ be
controlled in other circumstances, for example, when paining a sunset.

So perceptions are not controlled, only perceptual variables are
controlled. I do not control my perception of Mt. Everest, I may control my
perception of the speed at which Mt. Everest is approaching me.