Where's the dolphin? (was : Control Theory Entry in APA Dictionary of Psychology)

[From Rupert Young (2015.03.25 20.30)]

(Rick Marken (2015.03.23.1615]

Isn't the position what is varied as part of the output function in

order to control the perception? If your perception were disturbed,
by a pair of wonky glasses perhaps, then you would change your
position to compensate. We may be changing aspects of the
environment but not necessarily controlling them.

For there to be an environmental correlate of a perception there

must be something in the environment that changes along with the
perception. But we can control perceptions without anything in the
environment changing. For example, when viewing random dot
stereograms we can control a perception of a dolphin, say, simply
by changing the focus of our eyes. Where’s the environmental
correlate of the dolphin?

Noone is denying that there is an environment out there, just that

we are (always) controlling (rather than changing) aspects of the
environment or that there need be environmental correlates of the
perception.

Regards,

Rupert
···

RM: Perceptions always exist only internally. But there
is always an environmental correlate to a perception. In
your example, the environmental correlate is the physical
location of the trees and your position relative to them.

          RM: I think it's hard to deny that there is an

environment out there made up of variables from which our
perceptions are constructed by our perceptual functions.
So when we control perceptions we are always controlling
an aspect of the environment on which those
perceptions depend. Because of this it is possible for an
outside observer, using the principles of the test for the
controlled variable, to determine what perception(s)
someone else is controlling.

[Martin Taylor 2015.03.26.10.39]

[From Rupert Young (2015.03.25 20.30)]

  (Rick Marken (2015.03.23.1615]
  Isn't the position what is varied as part of the output function

in order to control the perception? If your perception were
disturbed, by a pair of wonky glasses perhaps, then you would
change your position to compensate. We may be changing aspects
of the environment but not necessarily controlling them.

What do you consider to be the environment of the ECU that controls

the relationship perception? (As a reminder, a basic Elementary
Control Unit contains a Perceptual Input Function, a Reference Input
Function, a comparator, and an Output Function.) To me, everything
else in the feedback loop is in the environment. That includes the
placement of the body.

But maybe you are raising a different issue, and arguing that a

relationship is not an aspect of the environment. That’s a trickier
question. The generic approach is that everything we perceive in the
environment is determined by (as Rick says) the basic physical
entities (perhaps quarks, gluons, and the like) and how they affect
our sensors and neural processing until they form a perception.
Taking this generic view, either everything we perceive corresponds
to some environmental correlate or nothing does. There’s no more
“reality” to a chair we can sit on than there is to a line-up of our
head and two trees. Either both are environmental correlates of
perceptions or neither is. Take your pick, but I think it is easier
to discuss PCT if we take both to be.

And to anticipate a possible further counter-example, so is your

perception of the honesty of your Senator, or of the probable future
direction of Labour Party policies in the UK. All depend on the
present and past dispositions of what has influenced your senses and
your genetic and epigenetic inheritance. Either they are all
environmental correlates of perception or none are.

  For there to be an environmental correlate of a perception there

must be something in the environment that changes along with the
perception. But we can control perceptions without anything in the
environment changing. For example, when viewing random dot
stereograms we can control a perception of a dolphin, say, simply
by changing the focus of our eyes. Where’s the environmental
correlate of the dolphin?

In the specific arrangement of the random dot patterns, and in the

momentary focal properties of your eyes, which alter the functional
relationships between the dot patterns and the perceptual function.
All of those are in the environment of the system that controls
whether you see a dolphin. (Incidentally, I don’t think focus
affects the random stereogram perception very much, unless the
defocus is extreme, but I’m taking your example as a statement of
hypothetical fact).

In this example, and possibly in the previous one, you seem to be

mixing up levels of perception. Perception of the alignment relation
between your head and the trees is not at the same level as
perception of the position of your head, nor your perception of each
tree individually. Control of your focus is not at the same level as
control of your perception of the dolphin.

  Noone is denying that there is an environment out there, just that

we are (always) controlling (rather than changing) aspects of the
environment or that there need be environmental correlates of the
perception.

I'm pretty closely aligned with Rick on this one, though as you can

probably tell from recent interchanges, I don’t completely agree
with his last sentence.

Martin
···

RM: Perceptions always exist only internally. But
there is always an environmental correlate to a
perception. In your example, the environmental correlate
is the physical location of the trees and your position
relative to them.

            RM: I think it's hard to deny that there is an

environment out there made up of variables from which
our perceptions are constructed by our perceptual
functions. So when we control perceptions we are always
controlling an aspect of the environment on
which those perceptions depend. Because of this it is
possible for an outside observer, using the principles
of the test for the controlled variable, to determine
what perception(s) someone else is controlling.

[From Rick Marken (2015.03.26.0940)]

···

Rupert Young (2015.03.25 20.30)

RY: For there to be an environmental correlate of a perception there

must be something in the environment that changes along with the
perception. But we can control perceptions without anything in the
environment changing. For example, when viewing random dot
stereograms we can control a perception of a dolphin, say, simply
by changing the focus of our eyes. Where’s the environmental
correlate of the dolphin?

RM: I should have said that there is an environmental basis of every perception rather than an environmental correlate. In the stereogram example the environmental basis of dolphin perception is the disparity between the dots in the stereo pairs. The appropriate stereopsis that produces the image is brought about by changing the convergence of the eyes, which is the output of the stereopsis control system. The perception of the dolphin, therefore, depends on the nature of the environment – the disparities in the stereogram – and the output of the control system; that is, as we say in PCT, p = f(d,o), where d is the environment (disturbances, the fixed binocular disparities between dots) and o is the system output (binocular convergence). It’s the same situation with the tree alignment example; the tree locations are fixed but the perception of the angular difference between the trees depends on their physical location (a fixed “disturbance”, d) and the observer’s location relative to the trees (the system output, o).

RY: Noone is denying that there is an environment out there, just that

we are (always) controlling (rather than changing) aspects of the
environment or that there need be environmental correlates of the
perception.

RM: My main concern in this discussion, as one who studies rather than builds control systems, is simply that we agree that perceptions always have an environmental basis; perceptions are always a function of environmental variables (actually, of the sensory effects thereof) and these functions define the aspects of the environment that are being controlled. This is a basic assumption of the PCT model, which says that p = f(d,o), where d is the environment and o is system output, the environmental effect of the system on the perception. So both o and d are environmental (ie. publicly accessible) variables.

RM: This aspect of the PCT model is important to a researcher because it means that it is possible to determine what perceptions a system is controlling using the test for the controlled variable. If perceptions did not have an environmental basis – if it were not the case that p = f(d,o) – then the study of living control systems would be impossible, which it is clearly not the case since I’ve been doing it successfully for 30+ years.

Best

Rick


Richard S. Marken, Ph.D.
Author of Doing Research on Purpose.
Now available from Amazon or Barnes & Noble

          RM: Perceptions always exist only internally. But there

is always an environmental correlate to a perception. In
your example, the environmental correlate is the physical
location of the trees and your position relative to them.

[From Rupert Young (2015.03.26 21.45)]

(Martin Taylor 2015.03.26.10.39]
  MT:

But maybe you are raising a different issue, and arguing that a
relationship is not an aspect of the environment. That’s a
trickier question. The generic approach is that everything we
perceive in the environment is determined by (as Rick says) the
basic physical entities (perhaps quarks, gluons, and the like) and
how they affect our sensors and neural processing until they form
a perception. Taking this generic view, either everything we
perceive corresponds to some environmental correlate or nothing
does. There’s no more “reality” to a chair we can sit on than
there is to a line-up of our head and two trees. Either both are
environmental correlates of perceptions or neither is. Take your
pick, but I think it is easier to discuss PCT if we take both to
be.

  MT: And to anticipate a possible further counter-example, so is

your perception of the honesty of your Senator, or of the probable
future direction of Labour Party policies in the UK. All depend on
the present and past dispositions of what has influenced your
senses and your genetic and epigenetic inheritance. Either they
are all environmental correlates of perception or none are.

RY: I think we are talking at cross purposes. Perceptions may depend

upon environmental variables but not necessarily correlate with
them.

(Rick Marken (2015.03.26.0940)
    RY: For there to be an

environmental correlate of a perception there must be something
in the environment that changes along with the perception. But
we can control perceptions without anything in the environment
changing. For example, when viewing random dot stereograms we
can control a perception of a dolphin, say, simply by changing
the focus of our eyes. Where’s the environmental correlate of
the dolphin?

  RM: I should have said that there is an environmental _basis_

of every perception rather than an environmental correlate. In
the stereogram example the environmental basis of dolphin
perception is the disparity between the dots in the stereo pairs.
The appropriate stereopsis that produces the image is brought
about by changing the convergence of the eyes, which is the output
of the stereopsis control system. The perception of the dolphin,
therefore, depends on the nature of the environment – the
disparities in the stereogram – and the output of the control
system; that is, as we say in PCT, p = f(d,o), where d is the
environment (disturbances, the fixed binocular disparities between
dots) and o is the system output (binocular convergence). It’s the
same situation with the tree alignment example; the tree locations
are fixed but the perception of the angular difference between the
trees depends on their physical location (a fixed “disturbance”,
d) and the observer’s location relative to the trees (the system
output, o).

  RY: Yes, it is the use of the word "correlate" that I was

questioning. In the eye the opening and closing of the iris
controls the light on the retina. That light perception depends
upon the light in the environment (the photons) but doesn’t
correlate with it. In fact as the light in the environment changes
the perception remains the same (due to perceptual control), so
there is no correlation between the two. This, I would have
thought, is one of the central tenets of PCT.

  However, I would also question whether there is an environmental

basis of every perception. As we go up the hierarchy perceptions
become more abstract and subjective and less related to objective
reality. If imagination, thoughts and beliefs are perceptions then
do these always have an environmental basis? How about unicorns,
fear and gods?

  Regards,

  Rupert

[Martin Taylor 2015.03.28.11.15]

[From Rupert Young (2015.03.26 21.45)]

  (Martin Taylor 2015.03.26.10.39]
    MT:

But maybe you are raising a different issue, and arguing that a
relationship is not an aspect of the environment. That’s a
trickier question. The generic approach is that everything we
perceive in the environment is determined by (as Rick says) the
basic physical entities (perhaps quarks, gluons, and the like)
and how they affect our sensors and neural processing until they
form a perception. Taking this generic view, either everything
we perceive corresponds to some environmental correlate or
nothing does. There’s no more “reality” to a chair we can sit on
than there is to a line-up of our head and two trees. Either
both are environmental correlates of perceptions or neither is.
Take your pick, but I think it is easier to discuss PCT if we
take both to be.

    MT: And to anticipate a possible further counter-example, so is

your perception of the honesty of your Senator, or of the
probable future direction of Labour Party policies in the UK.
All depend on the present and past dispositions of what has
influenced your senses and your genetic and epigenetic
inheritance. Either they are all environmental correlates of
perception or none are.

  RY: I think we are talking at cross purposes. Perceptions may

depend upon environmental variables but not necessarily correlate
with them.

Do you mean that one can't compute a Pearson correlation coefficient

between them? If you mean that, then you misunderstand me. I
intended “correlate” in the much more everyday sense of “if one
changes than the other is likely to”.

My point was that in your question [From Rupert Young (2015.03.25

20.30)]

  Isn't the position what is varied as part of the output function

in order to control the perception? If your perception were
disturbed, by a pair of wonky glasses perhaps, then you would
change your position to compensate. We may be changing aspects
of the environment but not necessarily controlling them.

you took the environment of the relationship control unit to begin

outside the skin of the person, whereas it actually begins at the
boundary between the relationship ECU and the level below. The
environment of an ECU is the area of the question mark below (and
above the refernce input, if you want).

![ECU.jpg|808x510](upload://rxwqYDIyR1kiF2csUzKEqCbvlsE.jpeg)

Supporting the idea that this is where the problem arises, you also

said:

  RY: Yes, it is the use of the word

“correlate” that I was questioning. In the eye the opening and
closing of the iris controls the light on the retina. That light
perception depends upon the light in the environment (the photons)
but doesn’t correlate with it. In fact as the light in the
environment changes the perception remains the same (due to
perceptual control), so there is no correlation between the two.
This, I would have thought, is one of the central tenets of PCT.

Part of the environment of the light perception is the iris, which

is on the path of the photon stream between whatever happens in the
outer world and the sensors (rods and cones). The other aspect of
the issue, which this example exposes, is that it’s not just the
current state of the environment but its history. Adaptation is a
memory process, and I think most perceptual functions produce
outputs that are highly dependent on the contrast between their
current inputs and their recent history (where “recent” depends on
the level of the perception, being measured in seconds at the level
of sensors, but perhaps months or years at the level of systems.

However, there was another possibility for why you asked the

question, so I asked whether you were not worrying about that, but
were instead suggesting that a relationship is not part of the
environment. To which I said [Martin Taylor 2015.03.26.10.39]:

  The generic approach is that everything we

perceive in the environment is determined by (as Rick says) the
basic physical entities (perhaps quarks, gluons, and the like) and
how they affect our sensors and neural processing until they form
a perception. Taking this generic view, either everything we
perceive corresponds to some environmental correlate or nothing
does. There’s no more “reality” to a chair we can sit on than
there is to a line-up of our head and two trees. Either both are
environmental correlates of perceptions or neither is. Take your
pick, but I think it is easier to discuss PCT if we take both to
be.

With the understanding that "correlate" could be replaced by

“coupled with” or something like that, does my comment now seem less
like talking at cross purposes?

Martin
···

RM: Perceptions always exist only internally. But
there is always an environmental correlate to a
perception. In your example, the environmental correlate
is the physical location of the trees and your position
relative to them.

[From Rupert Young (2015.03.29 18.30)]

(Martin Taylor 2015.03.28.11.15]

The generic approach is that everything we perceive in the environment is determined by (as Rick says) the basic physical entities (perhaps quarks, gluons, and the like) and how they affect our sensors and neural processing until they form a perception. Taking this generic view, either everything we perceive corresponds to some environmental correlate or nothing does. There's no more "reality" to a chair we can sit on than there is to a line-up of our head and two trees. Either both are environmental correlates of perceptions or neither is. Take your pick, but I think it is easier to discuss PCT if we take both to be.

With the understanding that "correlate" could be replaced by "coupled with" or something like that, does my comment now seem less like talking at cross purposes?

Well, I think that was the point I was making. And as I said in the preceding post you can have perceptions that have no "coupling" with environmental (external to individual) variables. For example, fear perception. One person might experience fear walking down a dark street late at night, another person may not.

Regards,
Rupert

[Martin Taylor 2015.03.29.12.37]

[From Rupert Young (2015.03.29 18.30)]

(Martin Taylor 2015.03.28.11.15]

The generic approach is that everything we perceive in the environment is determined by (as Rick says) the basic physical entities (perhaps quarks, gluons, and the like) and how they affect our sensors and neural processing until they form a perception. Taking this generic view, either everything we perceive corresponds to some environmental correlate or nothing does. There's no more "reality" to a chair we can sit on than there is to a line-up of our head and two trees. Either both are environmental correlates of perceptions or neither is. Take your pick, but I think it is easier to discuss PCT if we take both to be.

With the understanding that "correlate" could be replaced by "coupled with" or something like that, does my comment now seem less like talking at cross purposes?

Well, I think that was the point I was making. And as I said in the preceding post you can have perceptions that have no "coupling" with environmental (external to individual) variables. For example, fear perception. One person might experience fear walking down a dark street late at night, another person may not.

Good example of a situation in which memory and current environmental variables both contribute to a current perception. But I'd dispute "no coupling", because if the same street were well lit or you were accompanied by two sturdy police friends, the level of the "fear" perception would be lessened for the fearful person, or if partly hidden shadows of people were in the picture the level of fear perception would be increased for the brave person. Those would change only the current environmental input, not the memory component.

To me "coupling" doesn't mean "exclusively dependent on" or "a function only of". It means one varies when the other does, just as Z = X+Y varies when you change X, but cannot be determined from the value of X. I would say that the value of Z is "coupled" with that of X (and of Y), but is a function of X and Y, not of X nor of Y. That's also the sense in which I used "correlated" earlier.

Martin

[From Rick Marken (2015.03.30.0910)]

···

Martin Taylor (2015.03.29.12.37)–

RY: Well, I think that was the point I was making. And as I said in the preceding post you can have perceptions that have no “coupling” with environmental (external to individual) variables. For example, fear perception. One person might experience fear walking down a dark street late at night, another person may not.

MT: Good example of a situation in which memory and current environmental variables both contribute to a current perception.


Richard S. Marken, Ph.D.
Author of Doing Research on Purpose.
Now available from Amazon or Barnes & Noble

RM: Memory may be involved but I don’t think it actually “contributes” to the perception of fear. Fear (in PCT) is a label (a category perception) that is based on present-time perceptions, such as the perception of the physiological effects of adrenalin secretion and the fact that you are walking down a dark street at night. The same combination of perceptions may be labeled “excitement” rather than “fear” by another person.

RM: Memory comes into this only as a reference for a perception. People who imagine bad things happening when they walk down dark streets at night will set a reference for not having that perception of walking down such a street. So when they find themselves walking down a dark street at night there is an error which is not being corrected (unless they get out of there) and persistent uncorrected error leads to adrenalin secretion and the perception of the effects of the adrenalin combined with the perception of the fact that you are walking on a dark street at night results in your categorizing the perceived effects of the adrenalin as “fear”. But, again, the role of memory in this scenario is in the reference for perception, not in the perception itself. At least, in the PCT model that’s the case.

Best

Rick

[From Rupert Young (2015.04.02 21.00)]

(Martin Taylor 2015.03.29.12.37]

Well, I think that was the point I was making. And as I said in the preceding post you can have perceptions that have no "coupling" with environmental (external to individual) variables. For example, fear perception. One person might experience fear walking down a dark street late at night, another person may not.

Good example of a situation in which memory and current environmental variables both contribute to a current perception. But I'd dispute "no coupling", because if the same street were well lit or you were accompanied by two sturdy police friends, the level of the "fear" perception would be lessened for the fearful person, or if partly hidden shadows of people were in the picture the level of fear perception would be increased for the brave person. Those would change only the current environmental input, not the memory component.

This sounds like stimulus/repsonse to me.

Can you think of any examples where current environmental variables make no contribution to a current perception?

Rupert

[From Rupert Young (2015.04.02 21.10)]

(Rick Marken (2015.03.30.0910)]

This sounds like stimulus/response too. It seems to imply that
present-time perceptions (the environment?) causes, or is
responsible, for what we perceive.
How can different perceptions (category labels) be derived from the
same present-time perceptions without contribution from internal
processes (memory)?
There still seems to be something missing here (in my head anyway).
If memory (internal perspective) has no role in the perception then
why do different people perceive the same things in different ways?
Regards,
Rupert

···
        MT: Good example of a situation in

which memory and current environmental variables both
contribute to a current perception.

        RM: Memory may be involved but I

don’t think it actually “contributes” to the perception of
fear. Fear (in PCT) is a label (a category perception) that
is based on present-time perceptions, such as the perception
of the physiological effects of adrenalin secretion and the
fact that you are walking down a dark street at night.

        The same combination of perceptions

may be labeled “excitement” rather than “fear” by another
person.

        RM: Memory comes into this only as a

reference for a perception. People who imagine bad things
happening when they walk down dark streets at night will set
a reference for not having that perception of walking down
such a street. So when they find themselves walking down a
dark street at night there is an error which is not being
corrected (unless they get out of there) and persistent
uncorrected error leads to adrenalin secretion and the
perception of the effects of the adrenalin combined with the
perception of the fact that you are walking on a dark street
at night results in your categorizing the perceived effects
of the adrenalin as “fear”. But, again, the role of memory
in this scenario is in the reference for perception, not in
the perception itself. At least, in the PCT model that’s the
case.

[From Rick Marken (2015.04.02.2040)]

···

Rupert Young (2015.04.02 21.10)–

RY: This sounds like stimulus/response too. It seems to imply that

present-time perceptions (the environment?) causes, or is
responsible, for what we perceive.

RM: The model of perception in PCT is kind of S-R, where S is the input to the perceptual function and R is the perceptual signal output. In very simple terms the PCT model of perception is:

s.1 ----------------> | |

| |

s.2 ----------------> | |

| P(s.i) | -----------------> p

s.3 ----------------> | |

| |

s.n ----------------> | |

| |

RM: where the s.i are the sensory or lower level perceptual inputs to the perceptual function, P(s.i), and p is the perceptual signal output of the perceptual function. The perceptual function could compute p on the basis of all the s.i simultaneously; this would be how, for example, a configuration perceptual function probably works; the s.i would be sensations and P(s.i) might produce maximal perceptual signal output if the configuration of sensory inputs is a circle. say. So P(s,i) is a “circle detector”.

RM: But if P(s.i) is, say, a sequence detector (as in one of your robots) then P(s.i) would itself have to have memory since the relevant s.i are occurring sequentially rather than simultaneously. So if the s.i are circles of different sizes which occur one after the other and P(s.i) detects whether or not the sequence “small, medium, large” circle is occurring, P(s.i) it has to store up to two s.i to determine whether the third s.i is part of the sequence.

RM: So memory can certainly be part of a perceptual function itself; the inputs to the perceptual function have to be stored for a brief time in order to get a perception that is based on lower level perceptions/sensations that occur over time.

RM: What I don’t believe happens is that memory enters as an input to a perceptual function. That is, I don’t think perceptions are ever computed like this:

s.1 ----------------> | |

| |

s.2 ----------------> | |

| P(s.i, |

| m.1) | -----------------> p

s.n ----------------> | |

| |

m.1 ------> | |

^ | _______ |

|

|

| ___________________________

RM: where the memory input, m.1, to the perceptual function is a signal coming from inside the system, as a replayed reference signal (as per the memory model in B:CP). That is, I don’t believe perceptions are ever derived partially from non-sensory inputs like memory signal inputs. I think (though I would be happy to be proved wrong) that inputs to the perceptual function are always sensory based inputs; but the perceptual function itself may store some of the inputs for a short time in order to compute a perceptual signal based on sensory based inputs that vary over time.So, again, I believe that memory, in the form of temporary storage of sensory inputs, can be part of the computations done by a perceptual function; I don’t believe memory, in the form of traces of past experiences, can be the inputs to a perceptual function.

RY: How can different perceptions (category labels) be derived from the

same present-time perceptions without contribution from internal
processes (memory)?

RM: Perceptual functions at different levels of the hierarchy will compute different perceptions from the same set of sensory inputs. You gave a recent example in the form of random dot stereograms. The same pair of dot patterns viewed monocularly by each eye looks completely different when fused binocularly - a higher level perception of the same sensory input.

RY: There still seems to be something missing here (in my head anyway).

If memory (internal perspective) has no role in the perception then
why do different people perceive the same things in different ways?

RM: It could be different perceptual functions in the different people; Or the two people could be describing different levels of perceptions of the same thing. For example, we could both be listening to the same piece of music but you report hearing a particular melody and I report hearing a particular musical genre. You are reporting a sequence level perception and I’m reporting a category level perception of the same thing.

        RM:  Fear (in PCT) is a label (a category perception) that

is based on present-time perceptions, such as the perception
of the physiological effects of adrenalin secretion and the
fact that you are walking down a dark street at night.

        RM: The same combination of perceptions

may be labeled “excitement” rather than “fear” by another
person.

Best

Rick


Richard S. Marken, Ph.D.
Author of Doing Research on Purpose.
Now available from Amazon or Barnes & Noble

[From Rupert Young (2015.04.03 20.00)]

I think your reply only went to me so have included it below.

(Martin Taylor 2015.04.02.17.57]
Every perceptual function, every comparator, and every output function, by itself, IS a stimulus-response system. It's only the loop as a whole that isn't. The issue is whether any of the inputs to the perceptual function are from memory/imagination along with input from current environmental variables, and I still think your example is a good one.

Not necessarily, as far as the perception is concerned. If p = a + b, and a is increasing at the same rate as b is decreasing then p doesn't change, so no "response"; there is no environmental correlate of the perception.

Can you think of any examples where current environmental variables make no contribution to a current perception?

Yes, when I dream. Of course I can't be sure that no contribution to the perceptual function input comes from outside. I guess we can all remember having, or being told about, cases in which one has woken up feeling, say, cold, when one has just been dreaming of trudging over Arctic wastes, and things like that. But I doubt that most of the contents of a complex dream, such as the Baroque architecture of the houses I sometimes dream of, have much to do with current environmental variables (I've usually associated them with the complexities of trying to work out the arcane ramifications of PCT, but that's not current environmental variable variation).

How about when you're not dreaming? Here's a few suggestions, of perceptions that some people have that have no contribution from current environmental variables; that a Big Mac is a tasty meal, that America is the greatest country in the world, that women with short skirts are asking to be raped or that the world is 6,000 years old.

Regards,
Rupert

···

On 03/04/2015 00:05, Martin Taylor wrote:

[Martin Taylor 2015.04.02.17.57]

On 2015/04/2 2:58 PM, Rupert Young (rupert@perceptualrobots.com via > csgnet Mailing List) wrote:

[From Rupert Young (2015.04.02 21.00)]

(Martin Taylor 2015.03.29.12.37]

Well, I think that was the point I was making. And as I said in the preceding post you can have perceptions that have no "coupling" with environmental (external to individual) variables. For example, fear perception. One person might experience fear walking down a dark street late at night, another person may not.

Good example of a situation in which memory and current environmental variables both contribute to a current perception. But I'd dispute "no coupling", because if the same street were well lit or you were accompanied by two sturdy police friends, the level of the "fear" perception would be lessened for the fearful person, or if partly hidden shadows of people were in the picture the level of fear perception would be increased for the brave person. Those would change only the current environmental input, not the memory component.

This sounds like stimulus/repsonse to me.

Every perceptual function, every comparator, and every output function, by itself, IS a stimulus-response system. It's only the loop as a whole that isn't. The issue is whether any of the inputs to the perceptual function are from memory/imagination along with input from current environmental variables, and I still think your example is a good one.

Can you think of any examples where current environmental variables make no contribution to a current perception?

Yes, when I dream. Of course I can't be sure that no contribution to the perceptual function input comes from outside. I guess we can all remember having, or being told about, cases in which one has woken up feeling, say, cold, when one has just been dreaming of trudging over Arctic wastes, and things like that. But I doubt that most of the contents of a complex dream, such as the Baroque architecture of the houses I sometimes dream of, have much to do with current environmental variables (I've usually associated them with the complexities of trying to work out the arcane ramifications of PCT, but that's not current environmental variable variation).

Martin

[From Rupert Young (2015.04.03 20.30)]

(From Rick Marken (2015.04.02.2040)]

As per reply to Martin, this doesn't mean that there is an

environmental correlate of the perception. Perceptions, although
some may have an environmental basis, are entirely properties of
systems that construct perceptions, such as brains. Perceptions only
exist in brains; in the external world there are only primitive
variables. A tree falling in the forest makes no sound if there is
noone there to hear it.

I'm not sure I understand you, the two parts of this last sentence

seem to be saying the same thing. And my understanding of the figure
15.3 in B:CP is that memory goes up to form inputs to higher
perceptual functions.

I think I meant within different people; without individualistic

memory or learning.

Regards,

Rupert
···

RM: The model of perception in PCT is kind of S-R,
where S is the input to the perceptual function and R is
the perceptual signal output. In very simple terms the PCT
model of perception is:

s.1 ----------------> | |

| |

s.2 ----------------> | |

            |   P(s.i)   |

-----------------> p

s.3 ----------------> | |

| |

s.n ----------------> | |

| |

          RM: What I don't believe happens is that memory enters

as an input to a perceptual function. That is, I don’t
think perceptions are ever computed like this:

s.1 ----------------> | |

| |

s.2 ----------------> | |

| P(s.i, |

              |    m.1)   |

-----------------> p

s.n ----------------> | |

| |

m.1 ------> | |

^ | _______ |

|

|

| ___________________________

          RM: where the memory input, m.1, to the perceptual

function is a signal coming from inside the system, as a
replayed reference signal (as per the memory model in
B:CP). That is, I don’t believe perceptions are ever
derived partially from non-sensory inputs like memory
signal inputs. I think (though I would be happy to be
proved wrong) that inputs to the perceptual function are
always sensory based inputs; but the perceptual function
itself may store some of the inputs for a short time in
order to compute a perceptual signal based on sensory
based inputs that vary over time.So, again, I believe that
memory, in the form of temporary storage of sensory
inputs, can be part of the computations done by a
perceptual function; I don’t believe memory, in the form
of traces of past experiences, can be the inputs to a
perceptual function.

            RY: How can different perceptions (category

labels) be derived from the same present-time
perceptions without contribution from internal processes
(memory)?

          RM: Perceptual functions at different levels of the

hierarchy will compute different perceptions from the same
set of sensory inputs. You gave a recent example in the
form of random dot stereograms. The same pair of dot
patterns viewed monocularly by each eye looks completely
different when fused binocularly - a higher level
perception of the same sensory input.

                      RM: The same

combination of perceptions may be labeled
“excitement” rather than “fear” by another
person.

[From Rick Marken (2015.04.03.1200)]

···

Rupert Young (2015.04.03 20.30)–

RY: I'm not sure I understand you, the two parts of this last sentence

seem to be saying the same thing. And my understanding of the figure
15.3 in B:CP is that memory goes up to form inputs to higher
perceptual functions.

RM: You know what. You’re absolutely right. So PCT does allow for the possibility that imagined inputs – memory traces – could be a components of sense based perceptions.

RM: It would be interesting to see how this impacts control. I have no idea how to study this empirically but you could do it with a robot. Just have the robot controlling some perception that is partially derived from memory and see what happens. I don’t think it will be pretty;-)

Best

Rick

          RM: So, again, I believe that

memory, in the form of temporary storage of sensory
inputs, can be part of the computations done by a
perceptual function; I don’t believe memory, in the form
of traces of past experiences, can be the inputs to a
perceptual function.

Richard S. Marken, Ph.D.
Author of Doing Research on Purpose.
Now available from Amazon or Barnes & Noble

[Martin Taylor 2015.04.03.15.05]

[From Rupert Young (2015.04.03 20.00)]

I think your reply only went to me so have included it below.

(Martin Taylor 2015.04.02.17.57]
Every perceptual function, every comparator, and every output function, by itself, IS a stimulus-response system. It's only the loop as a whole that isn't. The issue is whether any of the inputs to the perceptual function are from memory/imagination along with input from current environmental variables, and I still think your example is a good one.

Not necessarily, as far as the perception is concerned. If p = a + b, and a is increasing at the same rate as b is decreasing then p doesn't change, so no "response"; there is no environmental correlate of the perception.

You are saying that if a + b doesn't change, p doesn't have a value???

Can you think of any examples where current environmental variables make no contribution to a current perception?

Yes, when I dream. Of course I can't be sure that no contribution to the perceptual function input comes from outside. I guess we can all remember having, or being told about, cases in which one has woken up feeling, say, cold, when one has just been dreaming of trudging over Arctic wastes, and things like that. But I doubt that most of the contents of a complex dream, such as the Baroque architecture of the houses I sometimes dream of, have much to do with current environmental variables (I've usually associated them with the complexities of trying to work out the arcane ramifications of PCT, but that's not current environmental variable variation).

How about when you're not dreaming? Here's a few suggestions, of perceptions that some people have that have no contribution from current environmental variables; that a Big Mac is a tasty meal, that America is the greatest country in the world, that women with short skirts are asking to be raped or that the world is 6,000 years old.

Good, but how are they relevent to the question that has been at issue, whether none, some, many, or all perceptions are based on inputes to the perceptual function from both memory/imagination and the current environment? Your examples and mine show that there can be perceptions without environmental input, but I don't think anyone familiar with B:CP would have disputed that.

To my mind, the remaining question is whether we ever have perceptions based only on the current values of environmental variables. I believe that doesn't happen, if only because all sensory systems adapt to the recent range of levels of input. For intensity variables, it's just called "adaptation". For location and configuration it's called "figural after-effect", and the prototype for velocity is the waterfall illusion. Having said that, I grant that subsequent processing might be able to recover the current values of environmental variables. Unlikely, but possible. Beyond these low-level effects, contrasts and asimmilations with what you have previously perceived generally influence what you now perceive based on sensory data.

Martin

···

On 2015/04/3 2:27 PM, Rupert Young (rupert@perceptualrobots.com via csgnet Mailing List) wrote:

[From Rick Marken (2015.04.03.1240)]

···

Martin Taylor (2015.04.03.15.05)–
MT: Your examples and mine show that there can be perceptions without environmental input, but I don’t think anyone familiar with B:CP would have disputed that.

RM: In PCT perceptions without environmental input are imaginations. So if those are the perceptions you are talking about when you refer to perceptions without environmental input then I don’t dispute it. But it you think there can be perceptions – the kind we are talking about when we talk about “control of perception” – without environment input, then I do dispute that.

MT: To my mind, the remaining question is whether we ever have perceptions based only on the current values of environmental variables. I believe that doesn’t happen,

RM: I agree, because many perceptions are based on environmental events that occur over time, like phonemic sequences that are the basis of the perception of words. But I think perceptions are always based, in large part if not entirely, on sensed effects of environmental variables. At least that is a basic assumption of PCT. It would be interesting to see if you have any evidence to the contrary.

Best

Rick

if only because all sensory systems adapt to the recent range of levels of input. For intensity variables, it’s just called “adaptation”. For location and configuration it’s called “figural after-effect”, and the prototype for velocity is the waterfall illusion. Having said that, I grant that subsequent processing might be able to recover the current values of environmental variables. Unlikely, but possible. Beyond these low-level effects, contrasts and asimmilations with what you have previously perceived generally influence what you now perceive based on sensory data.

Martin

Richard S. Marken, Ph.D.
Author of Doing Research on Purpose.
Now available from Amazon or Barnes & Noble

[Martin Taylor 2015.04.03.15.31]

My problem here is with "the same combination". Yes, it's possible,

because of the overlap on the “fight or flight” conditions. There
are conditions in which there’s no option, the “fight” or the
“flight” option wins the conflict. But there are intermediate states
in which the system history can leave the system on one or other
sheet of the fold catastrophe given the same input data. So one day
the person might feel “fear” while the next day it might be
“excitement”. I’m inclined to think, however, that “the same
combination of perceptions” is not actually the same, apart from the
part that is due to the current environmental input. Someone who had
once been mugged on a similar street might perceive it quite
differently from someone who has never encounter a violent stranger.
The one might see shadows of lurkers where the other sees only
interesting configurations of building shadows and reflections of
streetlights. The same might be true of two people, one of whom had
been told “I would never go there at night”, whereas the other had
been told “This is a lovely safe community”.
I have a specific real-life example of a difference in perception
based on prior experience. I was driving a couple of visitors from
New Jersey back to their hotel late one night, and we passed a
pretty girl walking down the street. One of them saw a prostitute
and said so. I was bewildered, because I didn’t see anything about
her to suggest such a profession. I saw only a girl walking
somewhere. He said that he could not imagine a pretty non-prostitute
walking alone in a city at night, so I prostitute was what he saw. I
don’t think he had a reference value for perceiving prostitutes any
more than I did. He just perceived differently because of his
experience.
Now this is where I disagree.
And the reason is that I do not see any reason why, given similar
reference values at higher levels (e.g. people of similarly
aggressive or placid nature), everyone would not set that same
reference value. Only if you have a reference value for wanting a
fight would you not. That’s quite different from having a perception
of the street as being safe or as being dangerous. If you perceive
it as safe, you won’t feel “fear”. If you perceive it as
“dangerous”, you might feel “fear” or “exhilaration”, but that’s not
the same as having a reference to perceive yourself as coming out
the other end unscathed.
In Rick’s personal PCT model, that’s the case, as we know. But
that’s not the same as the PCT model, for which there are many
possibilities.
Martin

···

I had decided not to get into another
interminable debate with Rick, and wasn’t going to respond. But at
the moment that control system is in strong conflict with one that
wants PCT to be analyzed correctly. I’m not saying that my
analysis is correct, but it’s different from Rick’s and its
unlikely (but possible) that they could both be correct.

[From Rick Marken (2015.03.30.0910)]

          Martin

Taylor (2015.03.29.12.37)–

                RY: Well, I think that was the point I was making.

And as I said in the preceding post you can have
perceptions that have no “coupling” with
environmental (external to individual) variables.
For example, fear perception. One person might
experience fear walking down a dark street late at
night, another person may not.

          MT: Good example of a situation in which memory and

current environmental variables both contribute to a
current perception.

        RM: Memory may be involved but I

don’t think it actually “contributes” to the perception of
fear. Fear (in PCT) is a label (a category perception) that
is based on present-time perceptions, such as the perception
of the physiological effects of adrenalin secretion and the
fact that you are walking down a dark street at night. The
same combination of perceptions may be labeled “excitement”
rather than “fear” by another person.

        RM: Memory comes into this only as a

reference for a perception.

        People who imagine bad things

happening when they walk down dark streets at night will set
a reference for not having that perception of walking down
such a street.

        ...But, again, the role of memory in

this scenario is in the reference for perception, not in the
perception itself. At least, in the PCT model that’s the
case.

(Martin Taylor 2015.04.03.15.05]

Not necessarily, as far as the perception is concerned. If p = a + b, and a is increasing at the same rate as b is decreasing then p doesn't change, so no "response"; there is no environmental correlate of the perception.

You are saying that if a + b doesn't change, p doesn't have a value???

No, as p = a + b. I am saying p doesn't respond to the change in the "stimuli".

How about when you're not dreaming? Here's a few suggestions, of perceptions that some people have that have no contribution from current environmental variables; that a Big Mac is a tasty meal, that America is the greatest country in the world, that women with short skirts are asking to be raped or that the world is 6,000 years old.

Good, but how are they relevent to the question that has been at issue, whether none, some, many, or all perceptions are based on inputes to the perceptual function from both memory/imagination and the current environment? Your examples and mine show that there can be perceptions without environmental input, but I don't think anyone familiar with B:CP would have disputed that.

We agree then. That the relevance is that there can be perceptions without environmental input. But this didn't seem to fit with all this talk of the control of environmental variables. The wider issue though is that we live (our experience) in a world of perceptions, which are not properties of the external world, but only "exist" within the nervous (or artificial) system. Hopefully that is not in dispute either.

To my mind, the remaining question is whether we ever have perceptions based only on the current values of environmental variables. I believe that doesn't happen, if only because all sensory systems adapt to the recent range of levels of input. For intensity variables, it's just called "adaptation". For location and configuration it's called "figural after-effect", and the prototype for velocity is the waterfall illusion. Having said that, I grant that subsequent processing might be able to recover the current values of environmental variables. Unlikely, but possible. Beyond these low-level effects, contrasts and asimmilations with what you have previously perceived generally influence what you now perceive based on sensory data.

Sounds logical.

Rupert

[From Rupert Young (2015.04.04 18.45)]

(From Rick Marken (2015.04.03.1200)]

RM: You know what. You're absolutely right. So PCT does allow for the possibility that imagined inputs -- memory traces -- could be a components of sense based perceptions.

RM: It would be interesting to see how this impacts control. I have no idea how to study this empirically but you could do it with a robot. Just have the robot controlling some perception that is partially derived from memory and see what happens. I don't think it will be pretty;-)

Maybe this is what is meant by the phrase "people see what they want to see".

Rupert

[Martin Taylor 2015.04.04.14.29]

(Martin Taylor 2015.04.03.15.05]

Not necessarily, as far as the perception is concerned. If p = a + b, and a is increasing at the same rate as b is decreasing then p doesn't change, so no "response"; there is no environmental correlate of the perception.

You are saying that if a + b doesn't change, p doesn't have a value???

No, as p = a + b. I am saying p doesn't respond to the change in the "stimuli".

Why should it be expected to, if the environmental state it perceives hasn't changed?

I don't see the point of your comment. In the two-line quote above that started this particular interchange, you seem to say that if something doesn't change it doesn't exist even though it can be perceived. The amount of money I have consists of some in a bank account ("a") and some in my wallet("b"). If I withdraw some from the bank and put it in my wallet, you seem to say that because the total ("p") doesn't change, I have no money even though I perceive that I do.

Martin

···

On 2015/04/4 12:36 PM, Rupert Young (rupert@perceptualrobots.com via csgnet Mailing List) wrote: