What's the problem?

[From Rick Marken (930514.1300)]

Oded Maler (930514.1900 ET) --

I'm having a little trouble figuring out what we are disagreeing
about. All I am saying is that everything we experience is
perception. No one has direct access to what is beyond their
senses. Models (which are also perceptions) are guesses about
what is 'out there' (past our senses) -- and some are pretty good
guesses given how well their behavior (a perception) corresponds
to our experience of the world (another perception). But we will
never know "for sure" what is really out there. All we know is that
something is out there -- and aspects of it's nature are captured
very well by some of our models.

Are you saying that some of our perceptions are better
representations of what is "out there" than others? Is this
what we are arguing about? For example, you say:

Do you think seriously the a
position of a line on a screen and a political situation are "equally
objective"?

My answer is "Yes". I was thinking of "objective" as meaning
"corresponding accurately to what is really 'out there'". In that
sense, these perceptions are equally objective -- meaning that
both are not objective AT ALL. They are both perceptions and the
degree to which either corresponds to what is actually out there is
equally unknown.

You may have thought I meant "objective" in the sense of "inter-
personal agreement". In this sense, lines and political situations
may not be equally "objective" -- but maybe not. Interpersonal
agreement just means that people agree on a description of what
they see. People may indeed be more likely to agree to call a line
a line (assuming they are all having the same perception) than to
call a beating a beating. But that's not what I meant. When you have
a perception of a line or a beating you have it -- it's a real thing.
Whether someone agrees or not that I am perceiving a computer
screen at the moment has little influence on that perception; it's
there for me, it's real (though I certainly would get a bit concerned
if someone came up and said "why are you looking at that bowl of
cherries? but the perception would remain. It's still a computer
screen to me). I can also look in the other office and have a perception
of a political situation -- two people are discussing the best way to
get funding for a particular project. I'm perceiveing what I perceive.
You might not perceive the same thing when you look at the same
scene -- but it would still be a perception for you. The perception --
whether you call it (and experience it as) a political discussion or a
friendly chat -- is undeniably there.

in the nice example that you
just gave, you see that the interpretation of the raw perception depends upon
the interpreter, the historical context, and a lot of subjective factors
that are absent in the "line on the screen" situation.

Maybe, maybe not. But in both cases there is a perception. Actually, even the
line on the screen perception can differ depending on "interpreter, historical
context, a lot of subjective factors"; the interpreter is the person
perceiving;
if s/he has no line detecting neurons there is no line (as there is apparently
no flies for a frog when the flies are not moving); the historical context is
everything the person knows about lines, computers and what the line means.
I've been in places where lines on screens could have meant the begining
of a nuclear holocaust. Pretty real,eh?; subjective factors just means the
aspects of the interpreter that result in the particular perception the
person is having; subjective factors made some of the lines in Tom's
cooperation experiment more important to some subjects than to others.

Relevance is a sufficiently vague term so that your rhethoric is not
completely false. But note what a quantum leap you have made from a
Hubel & Wiesel stuff about lines (whose position on the screen is
almost objective due to the concensus about it across most all human
subjects) than to a more speculative stuff about faceness (which is
higher and much less concensusial across individuals and cultures)
and than this unjustified generalization in all dimensions, letting
you perceive a global geo-political situation, something which
is many orders of magnitude larger, composed of many individual
and their collective percepotion.

Well, this is where I really need help in understanding what your point is.
My "quantum leap" was simply to say that line position, faceness and
geopolitical situations are perceptions. That just seems subjectively
obvious -- if you don't perecive them then how can you even talk
about them? The Hubel-Weisel discussion was just to show that there
is evidence of neural systems that COULD compute such perceptions.
It seems to me that you are trying to say that "global geopolitical
situations" cannot be perceptions. Is this what you are trying to
say? If so, how do you know they are not perceptions? Have you never
experienced a "global geopolitical situation"?

If all you are saying is that a person cannot perceive everything that is
actually involved in a perception like the one implied by words like "the
situation in Bosnia" (such as the feelings of every person involved, the
wants and desires of everyone involved, the terrain, household situation,
what everyone's having for dinner,etc etc)-- then I agree. But you also
cannot perceive everything involved in simpler perceptions like "line on the
screen" (you don't perceive the phosphers, the pixels, the photons, etc). All
I am saying is that, if you can talk about it, it's a perception. Whatever it
is
you think is important or relevant or historically interesting about anything
is a perception. Perceptions are not right or wrong -- though you can
perceive
things as right or wrong (relative to a reference) ; perception simply is what
you experience. What else is there?

So, what are we arguing about???

Best

Rick

[From Oded Maler (941007)]:

Rick Marken (941006.2030):

ยทยทยท

*
* >Bill Powers (941005.0605 MDT) said:
*
* >we can evaluate perceptions only in terms of other perceptions; there
* >is no way to establish an absolute correspondence between the
* >perceptual world and the physical world.
*
* Oded Maler (941006) replies:
*
* >I think this poses a substantial obstacle to the methodology of
* >"testing for the controlled variable". In fact, I think, it makes
* >it impossible, except for trivial artificial settings.
*

Rick:

* It's not an obstacle; the Test is quite possible (and has been done
* many times) in all kinds of non-trivial settings (including CSG-L).

[Noise deleted]

Suppose a researcher R wants to test the controlled variable of a
subject S. Suppose his conclusion is X. How is X expressed? Being
a perception it cannot be described in physical terms (maybe except
for simple sensations such as muscle tension). Being a perceptual
variable of S, it does not necessarily have a corresponding
perceptual varialble X' of R, such that R can use X' as an answer
(at least for himself). Hence the answer should be in terms of
the private and inaccessible perceptual world of S. Only under
the assumption of the existence of common shared perceptions ("square",
"distance of a cursor from a position in the screen"), and the assumption
that the answer should be drawn from this *fixed* set of perceptions,
testing can be done. I think this is a honest question.

* Are you guys trying to prove that we can't teach PCT or that you can't
* learn it?

It's all perception..

Best regards,

--Oded

--

Oded Maler, VERIMAG, Miniparc ZIRST, 38330 Montbonnot, France
Phone: 76909635 Fax: 76413620 e-mail: Oded.Maler@imag.fr

[From Rick Marken (941006.2030)]

Bill Powers (941005.0605 MDT) said:

we can evaluate perceptions only in terms of other perceptions; there
is no way to establish an absolute correspondence between the
perceptual world and the physical world.

Oded Maler (941006) replies:

I think this poses a substantial obstacle to the methodology of
"testing for the controlled variable". In fact, I think, it makes
it impossible, except for trivial artificial settings.

It's not an obstacle; the Test is quite possible (and has been done
many times) in all kinds of non-trivial settings (including CSG-L).

Hans Blom (941006) quotes himself:

assume, just for a minute, that there is no higher-level system and
that the highest-level reference level is fixed. Are the lower-level
reference levels now fixed as well? If not, might we not say that it is
the _perceptions_ that adjust the lower goals?

And observes that

This was not contradicted, and it keeps haunting me.

Be haunted no more. It's flat wrong. I hesitate to explain why because
it seems like it would be a complete waste of time. For the same
reason, I can't really bring myself to explain my answer to Oded. I can't
believe that after, what?, two years or more on CSG-L you guys are still
acting like you just came in contact with PCT for the first time, and
with a chip on your shoulders at that. What's the problem?

Are you guys trying to prove that we can't teach PCT or that you can't
learn it?

If Bill Powers answers either of these posts in his usual calm, patient,
helpful way, I'm calling the Pope myself (collect, of course) and giving
him a heads up for a future canonization possibility (I think it'll be OK;
Bill's not Jewish).

Best

Rick

Tom Bourbon [941007.0754]

[From Oded Maler (941007)]:

Rick Marken (941006.2030):
*
* >Bill Powers (941005.0605 MDT) said:
*
* >we can evaluate perceptions only in terms of other perceptions; there
* >is no way to establish an absolute correspondence between the
* >perceptual world and the physical world.
*
* Oded Maler (941006) replies:
*
* >I think this poses a substantial obstacle to the methodology of
* >"testing for the controlled variable". In fact, I think, it makes
* >it impossible, except for trivial artificial settings.
*

Rick:

* It's not an obstacle; the Test is quite possible (and has been done
* many times) in all kinds of non-trivial settings (including CSG-L).

Tom, now:
Oded, I believe that in Bill's post the important modifier was, "absolute:"
no _absolute_ correspondence between the perceptual and physical worlds.
That's just good old-fashioned psychophysics and for a more than a century it
hasn't stopped people from measuring all kinds of "psychophysical
functions" -- the performance-vs-stimulus functions that some people
mistakenly believe are ability-vs-stimulus functions.

When we do The Test we are in the same position as the psychophysicist,
looking at associations between our actions (introducing disturbances to
variables we believe the person may be controlling) and the actions of the
person (do they cancel the effects of our disturbance or not?) and drawing
inferences about the person's perceptions. Neither the psychophysicist nor
the PCTist can get beyond the position of monkeying around with the
environment and looking at the subject's behavior. My comments here also
apply to your following paragraph:

Suppose a researcher R wants to test the controlled variable of a
subject S. Suppose his conclusion is X. How is X expressed? Being
a perception it cannot be described in physical terms (maybe except
for simple sensations such as muscle tension). Being a perceptual
variable of S, it does not necessarily have a corresponding
perceptual varialble X' of R, such that R can use X' as an answer
(at least for himself). Hence the answer should be in terms of
the private and inaccessible perceptual world of S. Only under
the assumption of the existence of common shared perceptions ("square",
"distance of a cursor from a position in the screen"), and the assumption
that the answer should be drawn from this *fixed* set of perceptions,
testing can be done. I think this is a honest question.

See above. Also, see "Models and their
worlds," the obscure paper by Bill Powers and me in Closed Loop, and my
even more obscure paper in CL on "Mimicry, repetition and perceptual
control." In those papers, we begin to address the important point you
raise about possible limits on the set of perceptions the experimenter is
able to imagine a subject controlling. For example, in what looks like a
simple undisturbed pursuit tracking task with a target moving at uniform
speed back and forth between the same upper and lower positions on the
screen, an experimenter's first guess might be that the subject is using the
control handle to keep the cursor aligned with the target -- a perception
the experimenter can "share" with the subject. One way to test that idea is
to disturb the cursor and see if the person moves the handle so as to cancel
the effect of the disturbance and keep the cursor and target aligned: if
"yes," then perhaps the person is indeed controlling alignment of cursor and
target; if "no," perhaps cursor-target distance is not the controlled
perception. Always "perhaps."

If that test failed -- the person did not alter movements and keep cursor
and target aligned -- then perhaps alignment was not the controlled
perception. If the experimenter cannot imagine alternatives to that
hypothesis, the game is over. Are there alternatives? Yes. For one,
perhaps the person was making hand movements match target movements. One
way the experimenter can test for that idea is to change the target
movements and see if the person makes hand movements match the new pattern
If "yes" perhaps that is the controlled perception; if "no" (the person
kept on making the same movements as before, with neither the hand nor the
cursor following the target) probably not.

If the previous test did not work, perhaps the person was using a "motor
program, motor plan, motor schema," or some such set of
previously-established "commands" that cause the hand to move back and forth
to "track" a remembered set of hand positions that coincidentally follow the
original path of the target. To test that idea, we would need to apply
disturbances directly to the person's hand, moving it "out of position," as
we imagine that to be.

No one said The Test would always work or that it would always be easy. The
Tester must be prepared to use some imagination and to break free from easy
prior assumptions that everyone perceives the world the same way, or that
everyone controls the same perceptions.

I'll come back to some of this later, in a reply to some of Martin Taylor's
most recent comments about psychophysics.

Later,

Tom