motor programs, caused output, the curse

CSGnet Archive CSG1992
1

[Martin Taylor 920310]
(Rick Marken 920310)

and Martin Taylor (920309 16:30) says:

If
he detects on an algorithmic basis, his response will be determined by which
phoneme is more likely, or to which the given sound is more similar if it
falls outside the range of either real phoneme.

Motor programs? Responses determined by input probability?

Gary, help, the curse of PCT has taken over CSGNet.

Anybody want to talk controlled variables, maybe???

B'ain't nobody here but us percepts.

I was quoting something I wrote about 15 years ago in the passage you repeated.
But nevertheless, even though the viewpoint may change, it is still necessary
to have grounds for categorization before making the category judgment. Those
grounds do not change whether you are dealing with the control of a percept
or simply the part of the system between the sensors and the perceptual
input to a comparator.

In an ordinary psychophysical experiment, I think the main controlled percept
is that of experimenter satisfaction. That being stable, the experimenter
is able to induce responses from stimuli in ways we have often discussed.
One of them is to provide a stimulus pattern and ask whether it represents
class X or class Y, and to express increasing satisfaction as the agreement
between the subject's claim of X or Y agrees with the experimenter's opinion
of whether X or Y was presented (in other words, the subject does not perceive
high experimenter satisfaction by goofing off and answering X and Y
independently of what was presented).

I don't think you can simply throw away all of perceptual psychophysics because
the subject does not control what the experimenter presents. The controlled
percept is at a higher level, and the experimenter knows it. If you have
ever watched a naive subject in a psychophysical study, you will know how
apologetic they can be when they finish a run. They say something like "I'm
so sorry, but even though I tried, I'm sure I got lots of them wrong." It
is very hard to convince them that they can satisfy you by doing their best,
rather than by succeeding on every trial.

I would now only partially disavow the phrase "response determined by input
probability," in that it can be valid if there is some fixed reference against
which the percept is known to be compared. (Really there has to be a whole
set of references, but many of these are of the same class as Bruce has been
dealing with in talking about the socialization of linguistic norms. They
are tacitly assumed between experimenter and subject). At the time I wrote
the passage I quoted, none of those caveats would have occurred to me. Now
they would, but I would consider them part of the experimental environment
that justifies the treatment of the data as if the stimuli did determine
the responses. When the reference set deviates from what the experimenter
hopes it is, the effects are usually pretty obvious. In psychophysical
studies, there are usually lots of checks put in for that kind of problem,
even if the experimenters would not have described it in those terms.

Does that take the curse off, a teensy bit?

Next--habit patterns and motor programs?

Martin

2

[From Rick Marken (920310)]

Oded Maler (921003) says:

I believe that the case of syntax corresponds roughly to the variations
in the types of motor programs that control gait, etc.

and Avery Andrews says:

The idea of motor programs as a basis for syntax strikes me as quite
intriguing, but in a preliminary unguided foray into some of the
literature I did not get very far with it.

and Martin Taylor (920309 16:30) says:

If
he detects on an algorithmic basis, his response will be determined by which
phoneme is more likely, or to which the given sound is more similar if it
falls outside the range of either real phoneme.

Motor programs? Responses determined by input probability?

Gary, help, the curse of PCT has taken over CSGNet.

Anybody want to talk controlled variables, maybe???

Perplexed,

Rick

···

**************************************************************

Richard S. Marken USMail: 10459 Holman Ave
The Aerospace Corporation Los Angeles, CA 90024
Internet:marken@aerospace.aero.org
(310) 336-6214 (day)
(310) 474-0313 (evening)

3

[Rick Marken (920312)]

I said (Rick Marken 920310) in moderate jest:

Motor programs? Responses determined by input probability?
Gary, help, the curse of PCT has taken over CSGNet.
Anybody want to talk controlled variables, maybe???

And Martin Taylor (920310) replied:

I don't think you can simply throw away all of perceptual psychophysics
because
the subject does not control what the experimenter presents. The controlled
percept is at a higher level, and the experimenter knows it.

Well, I don't know that the experimenter really looks at it that way. I
used to do auditory psychophysics and I looked at the whole thing in
s-r terms. Yes, we did think about the subject's response goals (I used
SDT, of course). But even then we tried to see what was probably equivalent
to a reference signal in SDT (the criterion) as a response to input --
the payoff matrix. But I agree that we did undersatnd that the subject
was controlling (probably) at least one variable -- response probability or
(more likely) sequential response proabbility. I guess we also know that,
under "ordinary circumstances" (no monitary payoffs) the criterion setting
was secularly adjustable (just like a PCT reference signal).

I didn't mean to imply that this work was valueless. I was being a bit
sarcastic. But there was an element of seriousness there. I do think
that it is probably a waste of time to sift through the misconceptions
and fairly useless data of conventional psychology. I think it's just a
waste of time looking for areas of convergence between PCT and
conventional psychological research. . I think fair is fair --
conventional psychology is pretty content to ignore PCT; I think PCT
should just return the favor. I think we have already shown the
fundemental flaws in the conventional approach and given precise,
alternative explanations of what conventional psychology considers
to be some of its major phenomena (operant conditioning, coordinated
behavior, reflexes, etc). My personal feeling is that PCT people can be
much more productive by just starting from scratch, pretty much, rather
than trying to deal with a world of observations and theories that
were based on the wrong assumption about the nature of behavior. But that's
my personal opinion -- those of you who want to try to apply PCT as an
alternative to "conventional" models are free to try it; but as you begin
to get a deeper and deeper understanding of PCT I bet that you'll give
it up and just start studying control phenomena directly.

Does that take the curse off, a teensy bit?

Of course. Pay no attention to that loose canon behind the keyboard.

Hasta Luego

Rick

···

**************************************************************

Richard S. Marken USMail: 10459 Holman Ave
The Aerospace Corporation Los Angeles, CA 90024
Internet:marken@aerospace.aero.org
(310) 336-6214 (day)
(310) 474-0313 (evening)

4

[Martin Taylor 920316 11:10]
(Rick Marken 920312)

I seem to be reconnected again, but over the weekend there has been only this
one item from Rick and another from Bill. Is this right?

Rick says: I think fair is fair --

conventional psychology is pretty content to ignore PCT; I think PCT
should just return the favor.

One of the few things I was told in graduate school that has stuck with me as
something I still believe: If there are two schools of thought, each with
good reason claiming that they have the truth and the other doesn't, they
are probably both right, except in that claim.

It is obvious to me that the basic ideas of PCT have to be right, just as
are those of (say) signal detection theory or information theory. PCT cannot
work if the informational requirements are not met. The actions that a
subject performs in a signal-detection study cannot be accounted for without
PCT.

For some weeks I have been trying to get this question of zero references
in a stabilized hierarchy sorted out, so that I can get to my main point--that
most perceptual activity at any moment in time is passive and uncontrolled.
PCT has to acknowledge this--it is required by the informational arguments
and can't be wished away by faith or dogma.

Quite apart from that, it has to be the case that consistencies observed in
conventional psychological experiments tell us something about what goes on
inside a person, because consistencies represent something that resists
whatever (unknown) disturbances to which the subject is subject (!). It
may be, and usually is, hard to know what these consistencies tell us,
especially if we don't look at what the subject is trying to achieve (i.e. if
we ignore the PCT approach of looking at controlled perception). But the
difficulty of discovering what a poorly conceived experiment says should not
be sufficient reason for denying that it says anything useful. It may be
easier to regather the information in a better conceived experiment, but as
amply testified by the discussions on this net, designing such experiments
in PCT terms is not easy, either.

There's plenty of bath water, but there are babies in it, so be careful what
you throw away.

Martin

5

[Martin Taylor 920316 11:10]
(Rick Marken 920312)

I seem to be reconnected again, but over the weekend there has been only this
one item from Rick and another from Bill. Is this right?

Rick says: I think fair is fair --

conventional psychology is pretty content to ignore PCT; I think PCT
should just return the favor.

One of the few things I was told in graduate school that has stuck with me as
something I still believe: If there are two schools of thought, each with
good reason claiming that they have the truth and the other doesn't, they
are probably both right, except in that claim.

It is obvious to me that the basic ideas of PCT have to be right, just as
are those of (say) signal detection theory or information theory. PCT cannot
work if the informational requirements are not met. The actions that a
subject performs in a signal-detection study cannot be accounted for without
PCT.

For some weeks I have been trying to get this question of zero references
in a stabilized hierarchy sorted out, so that I can get to my main point--that
most perceptual activity at any moment in time is passive and uncontrolled.
PCT has to acknowledge this--it is required by the informational arguments
and can't be wished away by faith or dogma.

Quite apart from that, it has to be the case that consistencies observed in
conventional psychological experiments tell us something about what goes on
inside a person, because consistencies represent something that resists
whatever (unknown) disturbances to which the subject is subject (!). It
may be, and usually is, hard to know what these consistencies tell us,
especially if we don't look at what the subject is trying to achieve (i.e. if
we ignore the PCT approach of looking at controlled perception). But the
difficulty of discovering what a poorly conceived experiment says should not
be sufficient reason for denying that it says anything useful. It may be
easier to regather the information in a better conceived experiment, but as
amply testified by the discussions on this net, designing such experiments
in PCT terms is not easy, either.

There's plenty of bath water, but there are babies in it, so be careful what
you throw away.

Martin