Misquoting

From[Bill Williams 27 June 2004 12:30 PM CST]

[From Rick Marken (2004.06.27.0900)]

But it's important to keep in mind that Marc is not controlling for avoiding a tiger because he can't perceive the tiger. He is controlling for keeping >the sound of the growl at zero. If there is actually a tiger growling in the woods, then this control will lead to escape from the tiger as a fortuitous >side effect. If there actually is no tiger -- just someone making tiger sounds, say -- then this control will simply avoid the growl sound, and the >person making the sound will get a good laugh.

This sounds very much like what happened when I noticed that Rick had proposed that pedophiles would make good police officers and we all had a laugh.

Some people may think that "funny" is a perception rather than a really real reality. However, pragmatically, if the effect that this perception has is real, then the perception itself is also real.

Bill Williams

[From Bruce Gregory (2004.0627.1502)]

Rick Marken (2004.06.27.0900)

But it's important to keep in mind that Marc is not controlling for
avoiding a tiger because he can't perceive the tiger. He is
controlling for keeping the sound of the growl at zero. If there is
actually a tiger growling in the woods, then this control will lead to
escape from the tiger as a fortuitous side effect. If there actually
is no tiger -- just someone making tiger sounds, say -- then this
control will simply avoid the growl sound, and the person making the
sound will get a good laugh.

Please excuse me for intruding into a lengthy exchange that I can't pretend to understand. I do
think I understand Rick's position and I largely agree with it (I realize that is not worth much to
Rick or anyone else). I would, however, like to add a note to the above. Marc may think that he is
avoiding the tiger by running away from the growl. (In fact, I suspect he does.) So perhaps we
should distinguish between what Marc thinks he is doing and the perception he is likely to be
controlling to accomplish this goal. It may be overreaching at bit to say that Marc's avoiding the
tiger he imagines is there is "a fortuitous side effect." This is actually his intention.

Bruce Gregory

[From Bruce Nevin (2004.06.27 15:42 EDT)]

Marc Abrams (2004.06.27.0815 –

···

At 08:30 AM 6/27/2004 -0400, Marc Abrams wrote:

From
[

[From Bruce Nevin (2004.06.27
06:58 EDT)]

Thank you Bruce, a very nice PCT interpretation and
one I concur with. What I asked Rick for was an HPCT version of
that.

There can be no category level in a 1-level PCT model. The category level
as proposed in B:CP is only possible in an HPCT model.

    /Bruce

Nevin

[From Bruce Nevin (2004.06.27 16:05 EDT)]

Rick Marken (2004.06.27.0900)--

Bruce Nevin (2004.06.27 06:58 EDT)

I think the key difference is expressed in the phrase "but I know what a
tiger sounds like". Marc has a memory/reference for the sound of a tiger
growling that (according to the "but" in the story) you do not have.

This explanation is really no different than mine.

Sorry, you are mistaken, Rick. It is different because it depends on the
diversity of input to a category recognizer.

Regardless of why we think the growl is a tiger's or not, the important
part of the explanation (the part you leave out) is that you have to
assume that Marc _controls_ for the growl sound, acting to keep it at zero
by running from it, while I don't.

Correct, I left this out. But my unstated assumption was that you both
control for the presence of a tiger, and you both run on perceiving a tiger
nearby. Marc perceived a tiger nearby, just as on hearing a bark or seeing
a wagging tail I perceive a dog nearby, or on hearing the sound of a
housefinch singing I know that it is 12:00 o'clock because that is the
birdcall at that hour on the clock that my wife bought six years ago. You
did not perceive a tiger nearby because (according to the story, anyway)
you know there are no tigers in the woods of northern New Jersey, just as
you would perceive a bird singing nearby instead of recognizing that the
clock in the next room is announcing that it is noon (or midnight).

I said that this kind of synecdoche is a function of a category recognizer
according to B:CP. I just reminded myself that B:CP does not talk about a
category level. The notion of a category level was introduced later.

The fact that your explanation is different from the category-level
explanation is shown by the following:

Simply knowing that it's the sound of a tiger doesn't lead to any action.
There must be a control system in place, one that acts to control for the
sound of the growl. This would explain why Marc runs and I don't. Of
course, this explanation also predicts that Marc will run every time he
hears a tiger sound, whether it is produced by a nearby tiger, a person
imitating a tiger or a digital sound system or whatever. This means Marc
would be expected to run out of the room every time he hears a tiger on
the radio or on TV. Maybe he does.

This conclusion does not follow from the more complex explanation with
category recognition. The difference (in the story) between you and Marc is
not that he runs from the sound of a tiger and you do not. The assumption
is that you both run from the presence of a tiger, and the difference
between you is that he recognizes the presence of a tiger and you do not.

But it's important to keep in mind that Marc is not controlling for
avoiding a tiger because he can't perceive the tiger.

Here, quite plainly, is the difference between the two explanations.
According to the explanation in terms of a category recognizer, Marc does
perceive a tiger. The input is a growl. Probably there is some other input,
such as being in the woods, maybe even knowing there is a zoo nearby, but
that last was not mentioned in the story. For whatever reasons, the sound
of a tiger's growl is sufficient for Marc to perceive that a tiger is
present. On the other hand, you do not perceive a tiger. You perceive a
growling sound, but that perception does not enter the category recognizer
for "tiger", or, if it does, it is insufficient input for you to recognize
the presence of a tiger.

He is controlling for keeping the sound of the growl at zero.

He is controlling to avoid proximity to a tiger.

If there is actually a tiger growling in the woods, then this control will
lead to escape from the tiger as a fortuitous side effect. If there
actually is no tiger -- just someone making tiger sounds, say -- then this
control will simply avoid the growl sound, and the person making the sound
will get a good laugh.

Indeed. Tigers do have a characteristic sound, however. It would take some
experience and considerable practice to imitate it. He might say afterward
"Whoa! That was a good tiger imitation! How did you do that?"

         /Bruce Nevin

···

At 09:01 AM 6/27/2004 -0700, Rick Marken wrote:

From [Marc Abrams
(2004.06.27.1546)]

[From Bruce Nevin (2004.06.27 15:42 EDT)]

Sorry, I misread your post. If you’re suggesting
that your presumption of an existing ‘category’ level implies the
existence of the hierarchy as constructed in HPCT than I disagree with your
idea. If you are suggesting that there is some cognitive structure involved in
perceptions, than I agree with your assessment

Marc

Considering how often throughout history even intelligent
people have been proved to be wrong, it is amazing that there are still people
who are convinced that the only reason anyone could possibly say something
different from what they believe is stupidity or dishonesty.

Being smart is what keeps some people from being intelligent.

Thomas Sowell

Don’t argue with an idiot; people watching may not be able to
tell the difference.

Anon

I don’t
approve of political jokes. I’ve seen too many of them get elected

Anon

···

[From Bruce Gregory (2004.06.27.1625)]

Bruce Nevin (2004.06.27 16:05 EDT)

Here, quite plainly, is the difference between the two explanations.
According to the explanation in terms of a category recognizer, Marc does
perceive a tiger. The input is a growl. Probably there is some other input,
such as being in the woods, maybe even knowing there is a zoo nearby, but
that last was not mentioned in the story. For whatever reasons, the sound
of a tiger's growl is sufficient for Marc to perceive that a tiger is
present. On the other hand, you do not perceive a tiger. You perceive a
growling sound, but that perception does not enter the category recognizer
for "tiger", or, if it does, it is insufficient input for you to recognize
the presence of a tiger.

It seems to me that the term "category recognizer" is a bit glib. As is the claim that Marc perceives that a tiger is present. Marc perceives a growl. He infers that a tiger is present. Rick hears the growl but does not infer the presence of a tiger.

Bruce Gregory

There seems to be no shortage of people who value certainty above truth.

From [Marc Abrams (2004.06.27.1550)]

Glad you joined the fray BG.

[From Bruce Gregory (2004.0627.1502)]

should distinguish between what Marc thinks he is doing and the

perception

he is likely to be controlling to accomplish this goal.

An excellent point, what I perceive and what I control for are not
necessarily the same things.

I need to perceive something in order to control it (either consciously
or unconsciously), but I don't control all I perceive.

I see now why Rick might have thought it was a trick question, but it
was not intended to be. I was not asking what we might be controlling
for; I was asking what each of us perceived. My point was and is that we
all perceive different things and we take from the environment rather
than get from it. What we ultimately control for is gotten from our
perceptions, which include our imagination, and not solely from the
environment and our senses

This distinction I believe is an important one.

There is no difference between 'reality' and 'imagination' as far as
what I _perceive_. My actions to correct the error due to those
perceptions might make me aware that they are ineffective in reducing
the error and I might 'perceive' something differently because of that,
and change references because of it, but we are continually 'fooled' by
what we think we sense from the environment that turns out just not to
be so.

I would even go so far as to say we spend a lot of our time trying to
distinguish between what is 'real' and what isn't out there.

I'm not quite sure what the problem is here myself but thanks for
weighing in BG.

Marc

Considering how often throughout history even intelligent people have
been proved to be wrong, it is amazing that there are still people who
are convinced that the only reason anyone could possibly say something
different from what they believe is stupidity or dishonesty.

Being smart is what keeps some people from being intelligent.

Thomas Sowell

Don't argue with an idiot; people watching may not be able to tell the
difference.

Anon

I don't approve of political jokes. I've seen too many of them get
elected

Anon

[From Bill Powers (2004.06.27.1449 MDT)]

Bruce Gregory (2004.06.27.1625)--

It seems to me that the term "category recognizer" is a bit glib. As is
the claim that Marc perceives that a tiger is present. Marc perceives a
growl. He infers that a tiger is present. Rick hears the growl but does
not infer the presence of a tiger.

To say "a tiger" or "tigers" is to name a category. Tony the Tiger, the
image you used to see on cereal boxes, can be perceived as a familiar
configuration even without naming it or putting it into a category with
anything else, but "a tiger" or "tigers" refers to no particular
configuration: it is the name of a whole class of configurations containing
examples much too dissimilar ever to be mistaken for each other -- the
tiger in your tank, for example, or Dave Garroway's "Hello, old tiger," the
image of which does not resemble any other tiger.

The key is specificity. If you hear a growl and think it sounds like the
growl of that tiger you saw and heard on television half an hour ago, you
will be imagining _that_ configuration, _that particular_ animal they said
had just escaped from the zoo. That is not "tigers" or "a tiger", which has
no one shape.

Best,

Bill P.

[From Bruce Nevin (2004.06.27 17:16 EDT)]

Marc Abrams (2004.06.27.1546)–

Sorry, I misread your
post. If you re suggesting that your presumption of an existing category
level implies the existence of the hierarchy as constructed in HPCT than
I disagree with your idea. If you are suggesting that there is some
cognitive structure involved in perceptions, than I agree with your
assessment

Marc, I am only saying that the mechanism proposed for category
recognition is necessarily a part of HPCT. It requires inputs from other
levels, especially but not limited to lower “analogic” levels,
and a primary motivation for postulating it is to provide inputs to
higher “digital” levels. I am not claiming that the HPCT
account is correct or that the kind of mechanism that Bill has proposed
for category recognition is correct, that is, anatomically and
physiologically correct.

Your understanding of how category perception works may differ, and
probably does. Your rejection of Bill’s proposal about category
recognition is probably(I am guessing) one part of your rejection of
HPCT.

In other words, I make no claim that this account is the truth about
perception and control. I do claim that it is the truth about the
category-perception proposal and HPCT. If you’re saying that is what I am
wrong about, please clarify.

    /Bruce
···

At 04:17 PM 6/27/2004 -0400, Marc Abrams wrote:

[From Bruce Gregory (2004.0627.1732)]

Bill Powers (2004.06.27.1449 MDT)

To say "a tiger" or "tigers" is to name a category. Tony the Tiger, the
image you used to see on cereal boxes, can be perceived as a familiar
configuration even without naming it or putting it into a category with
anything else, but "a tiger" or "tigers" refers to no particular
configuration: it is the name of a whole class of configurations containing
examples much too dissimilar ever to be mistaken for each other -- the
tiger in your tank, for example, or Dave Garroway's "Hello, old tiger," the
image of which does not resemble any other tiger.

I was not disputing the existence of categories. They fall naturally out of any system that recognizes patterns. I was only suggesting that a "category recognizer" seems like a reification of the process of pattern recognition. I doubt there is a module in the brain that we will someday label "category recognizer." But of course I could be wrong.

Bruce Gregory

There seems to be no shortage of people who value certainty above truth.

[From Bruce Gregory (2004.0627.1748)]

Bruce Nevin (2004.06.27 17:16 EDT)

I am not claiming that the HPCT account is correct or that the kind of mechanism that Bill has proposed for category recognition is correct, that is, anatomically and physiologically correct.

I wonder if anyone would claim that quantum field is a correct description of physical reality? I mean is all of spacetime _really_ filled with quantum fields? Even Steven Weinberg, who once seemed to argue for this, has distanced himself from the picture of late. What would the harm be if HPCT accurately predicted the behavior of organisms but was not "anatomically and physiologically correct?" Would we discard the model? I think not. After all, Newtonian gravity is not "correct" but we seem to find uses for it every day.

Bruce Gregory

There seems to be no shortage of people who value certainty above truth.

From [Marc Abrams
(2004.06.27.1845)]

[From Bruce Nevin (2004.06.27 17:16 EDT)]

Marc, I am only saying that the mechanism proposed for category
recognition is necessarily a part of HPCT. It requires inputs from other
levels, especially but not limited to lower “analogic” levels, and a
primary motivation for postulating it is to provide inputs to higher
“digital” levels. I am not claiming that the HPCT account is correct
or that the kind of mechanism that Bill has proposed for category recognition
is correct, that is, anatomically and physiologically correct.

In other words, I make no claim that this account is the truth about
perception and control. I do claim that it is the truth about the
category-perception proposal and HPCT. If you’re saying that is what I am wrong
about, please clarify.

I agree with your assessment then. Let me add something
here. I am not opposed to the idea of a cognitive hierarchy or network in the
cortex or brain. In fact I like the idea. It’s the specifics, or as they say the devil is in
the details that trouble me about HPCT.

It really begins with the PCT definition of what a
perception is. All sensory modalities do not pass through the same anatomy nor
are they all processed the same way. To suggest that vision, hearing, taste and
smell all go through the same levels of construction is absurd. To also suggest
that imagination is some how ‘different’ than ‘sensory’
data is also, imho, not viable and I believe research bares this out. But that
is not the big issue.

Bill’s hierarchy was done through some serious
introspection. What wasn’t done before or since was collaboration with
sensory research to confirm the categorizations he made.

I strongly
suggest that anyone interested in the hierarchy re-read pgs 78-80 ‘A Look
Ahead’ in B:CP.

He states in B:CP starting on pg. 78

“The next eight chapters are concerned with an
attempt to construct a preliminary model of human behavioral organization as it
would exist after learning is complete
(italics are mine). …The simple structure I offer here appears to have
considerable explanatory power, but it should
be thought of only as a preliminary sketch made for the purpose of suggesting
research that will lead to further development of the model

(italics are mine)…”

So Bill and Rick, when did this ‘preliminary sketch’
get etched in stone? And what research made it happen?

When does ‘learning’ end? When does it begin? And
what does the hierarchy look like before
learning ends or before it begins?

After reading pages 78- 80 you will not wonder why I
question the hierarchy. I question it because Bill, in B:CP wanted it questioned. He also had hoped
that people would actually do some research to see, as he said, to help ‘fine
tune’ it. But instead, over the years the hierarchy turned into dogma as
Bill entrenched himself from his critics and built his bunker. I’m not
making this up kids. Read the book. And I am saying these things because it is
never too late to change and
rectify something as long as you have a breath in your body. But if you think
every criticism is aimed at ‘destroying’ you and your theory you’ll
never come out of your hole.

It’s unfortunate Bill that you can’t
distinguish between ‘friends’ of the theory and personal friends. You
don’t have to be one in order to be the other.

Marc

Considering how often throughout history even intelligent
people have been proved to be wrong, it is amazing that there are still people
who are convinced that the only reason anyone could possibly say something
different from what they believe is stupidity or dishonesty.

Being smart is what keeps some people from being intelligent.

Thomas Sowell

Don’t argue with an idiot; people watching may not be able to
tell the difference.

Anon

I don’t
approve of political jokes. I’ve seen too many of them get elected

Anon

···

from Mary Powers 2004.06.27

···

At 08:24 PM 6/26/2004, you wrote:

From [Marc Abrams (2004.06.26.2217)]

> From[Bill Williams 26 June 2004 8:00 PM CST]

Bill, I wasn't asking Rick to explain schizophrenia. I was asking him to
explain two aspects of it from an HPCT perspective. I was interested in
the processes of going into and out of hallucinations and the
'perceptions' that those hallucinations generated.

I guess I was hallucinating myself in thinking that maybe Rick had the
capacity to think outside the box and the current limited scope of HPCT.

Foolish me.

Marc

Marc, if you wonder where hallucinations fit into PCT, try looking at p.
278 in Living Control Systems and then think about why memory signals
entering the perceptual stream at a very low level are more real-seeming
that imaginings at higher levels.

Mary P.

From[Bill Williams

from Mary Powers 2004.06.27

From [Marc Abrams (2004.06.26.2217)]

> From[Bill Williams 26 June 2004 8:00 PM CST]

Bill, I wasn't asking Rick to explain schizophrenia. I was >>asking him to

explain two aspects of it from an HPCT perspective. I was >>interested in

the processes of going into and out of hallucinations and the

'perceptions' that those hallucinations generated.

I guess I was hallucinating myself in thinking that maybe Rick >>had the
capacity to think outside the box and the current limited scope >>of HPCT.

Foolish me.

Marc

Marc, if you wonder where hallucinations fit into PCT, try >looking at p.
278 in Living Control Systems and then think about why memory >signals
entering the perceptual stream at a very low level are more >real-seeming
that imaginings at higher levels.

Mary P.

Marc,

I think Mary Powers makes a valid point.

Now you can criticize me for defending what I perceive to be among the many defensible features of Bill Powers' application of control theory to human behavior.

Bill Williams

···

At 08:24 PM 6/26/2004, you wrote:

[From Rick Marken (2004.06.27.2250)]

Bruce Nevin (2004.06.27 16:05 EDT)

The fact that your explanation is different from the category-level
explanation is shown by the following:

Simply knowing that it's the sound of a tiger doesn't lead to any
action.
There must be a control system in place, one that acts to control for
the
sound of the growl. This would explain why Marc runs and I don't. ...

This conclusion does not follow from the more complex explanation with
category recognition. The difference (in the story) between you and
Marc is
not that he runs from the sound of a tiger and you do not. The
assumption
is that you both run from the presence of a tiger, and the difference
between you is that he recognizes the presence of a tiger and you do
not.

But in Marc's story I didn't run. The difference between your
explanation of what occurred and mine is that yours explains something
that didn't happen in the scenario described. Also, your category
level explanation of tiger "recognition" doesn't mean that a tiger is
actually out there.

But it's important to keep in mind that Marc is not controlling for
avoiding a tiger because he can't perceive the tiger.

Here, quite plainly, is the difference between the two explanations.
According to the explanation in terms of a category recognizer, Marc
does
perceive a tiger.

He perceives the categorical perception "tiger", which is basically the
word "tiger", not a specific tiger coming at him.

He is controlling to avoid proximity to a tiger.

He can't because he has no perception of proximity to the tiger. All he
has, according to your explanation, is a perception of the category
"tiger" which was triggered by combining the growl, the woods, etc in a
category-type perceptual input function that produced the output
category perception "tiger". In order to stop perceiving "tiger" Marc
has to remove all or some of the inputs to the category perceiver,
which he can do by moving away from the growl and out of the woods. In
the unlikely event that there happens to be an actual tiger in the
woods of NJ (as assumed in Marc's scenario) then controlling this
category perception will, as an accidental side effect, get Marc away
from the actual tiger. But if, as is far more likely, there is no tiger
out there but only an animal that sounds like a tiger through the woods
-- a boar or pig, say -- then getting rid of the tiger category
perception would not be getting Marc away from an actual tiger.

Regards

Rick

···

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

[From Bill Powers (2004.06.28.0255 MDT)]

Bruce Gregory (2004.0627.1732)--

I was not disputing the existence of categories. They fall naturally out
of any system that recognizes patterns. I was only suggesting that a
"category recognizer" seems like a reification of the process of pattern
recognition. I doubt there is a module in the brain that we will someday
label "category recognizer."

There are many forms of "pattern recognizer" in HPCT -- 11 forms right now,
to be exact. For shapes, we can have recognition at level three,
configurations, or at level 9 that I call the category level. At level
three, configurations are perceived but not named; furthermore, they are
explicit shapes, not classes of shapes. This dog is not that dog. At level
9, as I conceive it, categories are formed from sets of lower-level
perceptions, such that any perception belonging to the category will yield
the same perceptual signal indicating presence of the category. This
Pomeranian and that Irish Wolfhound are dogs. The items that belong to the
category can be very different; as I indicated in my example, Tony the
Tiger, or a real Siberian male tiger 20 feet from nose to tip of tail, or a
cute cub in a zoo, or an old injured and dying tiger posing no threat to
anyone, and so on through an infinite variety differences. In every case
we're talking about "a tiger."

"Pattern recognition" is so inclusive that I think it discards a lot of
useful distinctions, such as that between a spatial pattern and a temporal
pattern, or a pattern of mistakes or a pattern of symmetries. Perceiving
such things requires very different kinds of input functions, I should think.

Best,

Bill P.

[From Bruce Gregory (2004.0628.0700)]

Bill Powers (2004.06.28.0255 MDT)

"Pattern recognition" is so inclusive that I think it discards a lot of
useful distinctions, such as that between a spatial pattern and a temporal
pattern, or a pattern of mistakes or a pattern of symmetries. Perceiving
such things requires very different kinds of input functions, I should think.

I should think so too. But that does not, in my view, diminish the value of identifying a common mechanism. Much as the word "control" identifies a common mechanism with different kinds of input and output functions.

Bruce Gregory

There seems to be no shortage of people who value certainty above truth.

[From Bill POwers (2004.06.28.0815 MDT)]

Bruce Gregory (2004.0628.0700) --

But that does not, in my view, diminish the value of identifying a common
mechanism. Much as the word "control" identifies a common mechanism with
different kinds of input and output functions.

That's an interesting proposition, but I'd have to see it spelled out. Are
all the things we call "patterns" detected by the same kind of mechanism?
I'm not saying this isn't true, but I'd like to see a fuller discussion of
what constitutes a pattern, and what a detector of patterns would have to
be able to detect. Is what I call sequence perception the same as a
temporal pattern perceiver? Does a sensation perception consist of a
pattern of intensities? In other words, is there some fundamental mechanism
of pattern detection that could be repeated at one level after another to
generate diverse levels of perception such as those I have proposed, or
something like them?

My suspicion is that "pattern" is a verbal label that we apply to many
different experiences that are created by different mechanisms, the only
thing they have in common being that they are drawn from more than one
lower-order perception. As a parallel, consider the term "utility." Many
things are said to have this property, but it is only an expression of the
fact that we make use of many things -- not that they have anything in
common. Could that also be the case for the term pattern? We can apply this
term to many different perceptions, but does this mean the perceptions, or
the various means of deriving them, have anything in common?

I'm open to suggestions on this matter. If there's some underlying
principle of perception that could be applied at all levels, we might have
a hope of building an actual multilevel model of perception (which right
now is impossible). So I'm definitely interested, despite my doubts. What
have you got?

Best,

Bill P.

[From Bruce Nevin (2004.06.28 10:58 EDT)]

Bill Powers (2004.06.28.0255 MDT)--

There are many forms of "pattern recognizer" in HPCT -- 11 forms right now,
to be exact. For shapes, we can have recognition at level three,
configurations, or at level 9 that I call the category level.

You figure category control at level 9 now? Last I recall, you put it at
level 7, but were considering moving Relationship perception above it,
which would have made it level 6. Programs, Principles, and Systems were
the "subjective" levels 9, 10, and 11. What has happened in the interim?

At level
three, configurations are perceived but not named; furthermore, they are
explicit shapes, not classes of shapes. This dog is not that dog.

I know that flexure of an arm or of fingers into a fist is more or less of
one configuration. However, for perception of this dog from this point of
view and then from this other point of view to be "the same dog" is much
more complicated. Clearly, it is not a scalar value from "more dog" to
"less dog" in the way that limb flexure is more extension and less
extension. It is for this sort of thing (character recognition, face
recognition, etc.) that pattern recognition is typically invoked, as I recall.

         /Bruce Nevin

···

At 03:06 AM 6/28/2004 -0600, Bill Powers wrote:

[From Bruce Nevin (2004.06.28 11:30 EDT)]

Rick Marken (2004.06.27.2250)--

The fact that your explanation is different from the category-level
explanation is shown by the following:

Simply knowing that it's the sound of a tiger doesn't lead to any
action.
There must be a control system in place, one that acts to control for
the
sound of the growl. This would explain why Marc runs and I don't. ...

This conclusion does not follow from the more complex explanation with
category recognition. The difference (in the story) between you and
Marc is
not that he runs from the sound of a tiger and you do not. The
assumption
is that you both run from the presence of a tiger, and the difference
between you is that he recognizes the presence of a tiger and you do
not.

But in Marc's story I didn't run.

Exactly so, Rick. In your explanation, you didn't run because you don't run
from the sound of a growl but Marc does. In my explanation, you didn't run
because you didn't perceive a tiger but Marc did.

The difference between your
explanation of what occurred and mine is that yours explains something
that didn't happen in the scenario described.

I don't think you mean this, but -- your words do have a correct
interpretation: the thing that didn't happen was your running away.

Also, your category
level explanation of tiger "recognition" doesn't mean that a tiger is
actually out there.

Of course not. Nor did I say that it does. Any perception may be mistaken.
All that is relevant is that Marc perceives that there is a tiger out
there, and you do not.

But it's important to keep in mind that Marc is not controlling for
avoiding a tiger because he can't perceive the tiger.

Here, quite plainly, is the difference between the two explanations.
According to the explanation in terms of a category recognizer, Marc
does perceive a tiger.

He perceives the categorical perception "tiger", which is basically the
word "tiger", not a specific tiger coming at him.

The word "tiger" is (according to the hypothesis) one of the inputs to the
tiger-recognizer. The category perception is not "basically" the word any
more than it is "basically" any of the other inputs.

If I'm swimming in a river and someone says there are piranhas there, I get
out fast. I don't look around to perceive particular piranhas first.
Somebody says "cops!" and the criminals scatter and run. They are not
running from the word. Nor have they perceived a specific policeman coming
at them. This is a very general phenomenon. If it is not possible for the
model to explain control of a perceived category just as for a perceived
exemplar of the category, something needs fixing. I enter a dark room. I
hear a dog start growling. I don't respond as though someone had
matter-of-factly said the word "dog". I back out of the room and look for a
light. The specific, present growl suffices for me to perceive an entire,
specific dog.

Here's one way this could work. Marc hears a growl. The category input
function recognizes it, possibly together with other inputs, as a tiger.
The category recognizer puts systems that control its other inputs into
imagination mode. It is a specific, present growl. As such, it is a
perception of a specific, present tiger. The imagined perceptions are not
so specific (being cut off from their lower-level inputs) but otherwise
they are indistinguishable from perceptions of a specific, present tiger.
Marc runs.

This might not be so difficult if categorization were pervasive in the
model rather than the inputs having to go up to a higher level. That's a
discussion I've engaged in before, but I'm not up for it right now.

         /Bruce Nevin

···

At 10:53 PM 6/27/2004 -0700, Rick Marken wrote: