they already know that

[bnhpct]

[From: Bruce Nevin (Wed 92107 13:19:46)]

(Rick Marken (921007.0800) ) --

The "Hierarchical behavior of perception"
paper was rejected by "Theory & Psychology". The two reviews were the
nicest I have ever received. Basically, both said it was a good paper
but "we already know that" (that behavior is controlled perception).
I love it! So why are they still acting like they don't, I wonder?

Can you point out something they do that is incompatible with "already
knowing that"? A couple of f'rinstances, familiar and generally
accepted methods and results whose incompatibility with CT you can
succinctly demonstrate, and which are likely to fall within the
specializations of your reviewers? They might notice that what they are
agreeing to is not quite what you have said. Might not get the paper
published, but the reasons for rejection might change in interesting
ways. Incompatibility of the reasons for rejection, especially from the
same reviewers, might even provide some kind of leverage with the
editor.

The voice of inexperience . . .

  Bruce
  bn@bbn.com

[Human_Factors_Lab]

[From Rick Marken (921007.1230)]

Brice Nevin (921007) --

Can you point out something they do that is incompatible with "already
knowing that"?

Well, the "experimental method" as commonly used in psychology; they don't
test for controlled variables.

The paper was somewhat philosophical. I tried to make one big point
explicitly -- that performance limitations (in terms of speed of
behavior) may be limitations on the ability to perceive (rather than
to produce the actions that produce) the results that are being controlled.
I also tried to show how one can explore the perceptual (and, thus,
behavioral) hierarchy. There was a lot in this paper that I thought was
fairly new and interesting; alas, not the reviewers.

One would have to read the paper to get a good sense of what the reviewers
missed (or got right). The interesting thing is that both reviewers said
that they were "hard pressed" to find anything new in the paper; both
mentioned TOTE units (missing the fact that Miller et al never understood
that working versions of these "units" would control their perceptions),
they also mentioned people who had presumably worked on the relation between
perception and behavior (apparently ignoring the fact that the paper was
about the fact that, from the perspective of the HPCT model, behavior IS
perception). My efforts to make the paper more respectable by referring
to work in the accepted literature (and always in a friendly way) backfired
with the first reviewer who saw these references as old hat.

The second reviewer gets high marks for saying that he has followed my
research for several years with great interest (I know who the reviewer
is; I referred to several of his reearch findings in the paper). I just
wish he had followed it with a tad more understanding (though, fair's
fair, he seems to wish the same about me with respect to his research,
though all I did was use his results, not his interpretations thereof).

Anyway, the reviews were friendly. They didn't beg me to rewrite and
resubmit, but they were moderately encouraging. I might try to
rewrite it, but it seems like a topic that might be more interesting
to those who have already passed PCT 101. In other words, I'd rather
leave it as is and publish it in the Journal of Living Control Systems.
But we'll see; always nice to have a paper out there, waiting to be
rejected.

Hasta Luego

Rick

···

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

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

[Penni_Sibun]

(ps 921097.1400)

   [From Rick Marken (921007.1230)]

   >Can you point out something they do that is incompatible with "already
   >knowing that"?

   Well, the "experimental method" as commonly used in psychology; they don't
   test for controlled variables.

take an example from the psych lit.. show yr claim is true. show why
they should. show what they miss if they don't. if you can't come up
w/ a telling example of why experimentaly methodology is wrong, and
why your way fixes (some of) the problems, that's your problem, not
the reviewers'.

   The paper was somewhat philosophical. I tried to make one big point
   explicitly -- that performance limitations (in terms of speed of
   behavior) may be limitations on the ability to perceive (rather than
   to produce the actions that produce) the results that are being controlled.

that's a reasonable point, and i'm not surprised they didn't disagree.
but unless you have *some* (new) way of demonstrating it or arguing it
(in their paradigm, not yours), then you haven't in fact added to the
literature.

   I also tried to show how one can explore the perceptual (and, thus,
   behavioral) hierarchy. There was a lot in this paper that I thought was
   fairly new and interesting; alas, not the reviewers.

but it's not interesting if one doesn't agree w/ the underlying
assumptions and doesn't see how the discussion has any impact on one's
own assumptions. if you showed how well-known phenomena can be
analyzed differently and novelly in yr hierarchy--*that* way you can
make an *argument* for your hierarchy. but that won't work if the
reader must assume the hierarchy first and have no way to relate the
discussion to their own ways of looking at things.

   they also mentioned people who had presumably worked on the relation between
   perception and behavior

why ``presumably''? plenty of people have had things to say about
this.

  (apparently ignoring the fact that the paper was
   about the fact that, from the perspective of the HPCT model, behavior IS
   perception).

but see, i'm not convinced, from the kinds of things you write about,
that it's coherent for you to say ``perception is behavior.'' i think
i can map from your model straightforwardly into one w/ pieces labeled
``perception'' and ``behavior.'' you haven't convinced me i should
consider the terms synonymous or what that would mean.

cheers.

        --penni

[Human_Factors_Lab]

[From Rick Marken (921008.1300)]

penni sibun (921097.1400) --

Ah, it's nice to have you back.

if you showed how well-known phenomena can be
analyzed differently and novelly in yr hierarchy--*that* way you can
make an *argument* for your hierarchy.

Actually I did that, though I didn't do it quantitatively, unfortunately
(like by showing a working model) -- mainly because the right data wasn't
available and I was not interested in doing the experment at the time --
but I might do it eventually; this was really a theoretical paper, after
all. The "well known" phenomenon was really just a finger tapping study
done by Rosenbaum (described in JEP:HPP, 1984). He had subjects make
sequences of finger taps as rapidly as possible. The result of interest to
me was that the speed limit was about 4/sec, the same as the perecptual
limit for perceiving auditory and visual sequences. My alternative theory
of what was going on (Rosenbaum had an output generation model) was that
subjects were controlling a sequences of perceptions of finger tip
pressure. I suggested, in the paper, that one piece of evidence to support
this would come from a study the subjects' ability to perceive finger pressure
sequences that are presented to them (that are not generated by their
own taps) -- my guess is the same limit will be found. So the speed of
finger tip sequence production is limited by the ability to perceive
(and hence control) sequence -- not by limitations in the output
process. There are other experiments that I could think of -- the most
obvious would be to add disturbances to the finger movements and
see that this has little effect on the perceived sequences. But the
goal of the paper was to find evidence of perceptual control in the
existing literature -- not easy given that so much of the data is fairly
useless because it is averaged over people.

but see, i'm not convinced, from the kinds of things you write about,
that it's coherent for you to say ``perception is behavior.'' i think
i can map from your model straightforwardly into one w/ pieces labeled
``perception'' and ``behavior.'' you haven't convinced me i should
consider the terms synonymous or what that would mean.

Believe me, I understand your problem. I've been trying to explain
and demonstrate this stuff for years and it is not easy to communicate
what I consider to be obvious points. I'm sure it's as frustrating for
you as for me. I thought I did a pretty good job of explaining in the
paper that "behavior" refers to controlled results of action; actions
are the means by which behaviors (controlled results) are brought to
their intended levels. In a control loop, the result that is untimately
controlled is the perceptual input variable. The actions used to produce
this results are not "behaviors" from the point of view of the behaving
system but they may appear to be "behaviors" from the point of view
of the observer of the system, simply because they are produced by the
system (the word "behavior" is typically used to refer to any result of an
organisms muscle actions). For example, in the "rubber band" demo, the
position of the knot is a behavior of the subject; the position of the
subject's hand is NOT (though it would be seen as such by any observer
of the situation). The position of the knot is controlled -- it is an
intentional result of the subject's actions. The position of the hand
is not controlled -- it will be moved wherever necessary to keep the
knot of target.

I think you raise an interesting point about showing how PCT is a better
model of some "well known" phenomenon than some current explanation. One
problem we have in PCT is that most of the "well known" phenomena around
are, from our point of view, not phenomena at all. We went through this
exercise some time ago when we looked, in detail, at a study of voice
onset time in phoneme recognition. It turned out that the phenomenon was
only there statistically -- at best. Maybe you could pick a "well known"
phenomenon, give the reference to the reserach article it is described
in (so we can all get a copy of the results) and we can see what PCT has
to say about it.

Best regards

Rick

···

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

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