Statistics, Information, PopPCT, sci.cog

[From Rick Marken (940407.1030)]

Dan Miller (940406)

Do you not use statistics to indicate the magnitude, direction,
significance, and reliability of your observations?

Magnitude and direction, yes; significance and reliability, no. We use
summary statistics to describe the data. We use the inter-ocular trauma test
to decide on significance. The most important "significance test" in PCT is
seeing whether subject data matches model data closely. How close is close?
Well, it's like quality; you know it when you see it.

Martin Taylor (940407 09:50) --

It is the action of the loop AS A WHOLE that matters. But the action of the
loop as a whole CAN be analyzed by looking at the functions of its parts and
how they interrelate. That notion seems to terrify Officer Marken.

I work in a rough neighborhood; surrounded by interactionists,
dynamical systems theorists, cognitivists, and (scariest of all) information
theorists; I have to be on my guard -- Magnum at the ready.

The main problem I have in engaging in discussions publicly on CSG-L
is that Officer Marken persists only in trying to lock me into a cell
that has a view only of part of the control loop. I cannot work with
only this view, and sensible discussion from that cell is impossible to
continue. I therefore have decided to refrain from trying, in his presence.

I simply do not understand the reason for this "Persistence of Vision."

You keep throwing me off with all this talk about perceptual information,
a notion that makes no sense, formally or informally, in the context of PCT.
At best, it is an irrelevant diversion; simply a mistaken way of looking at
how the control model actually works. This is no problem; you could say that
control systems operate on the basis of liverwurst, for all I care. The
problem (as I have said before) is that "perceptual information" could be
used as a version of favorite fish -- a red herring. If people really thought
there was anything to this idea, then they might start looking for
the perceptual information that is used by control systems; they would do
research that was oriented in the wrong direction -- toward finding the
informational correlates of behavioral outputs, for example, or the rate at
which information is transported around the loop, and so on. In other words,
they could end up following the red-herring that most research psychologists
are already following; the one that suggests that behavior can be understood
in terms of the informational content of the environment.

I suppose it's not necessary for me to enforce the PCT laws against
perceptual information so strictly; after all, the best way to keep people
from following a red herring is to point to the real herring instead
(assuming they want the real one, and not the red one). So I will try
to ignore what I'm sure will be futuree violations of the perceptual
information code. As long as it doesn't divert you from the task of
identifying controlled perceptual variables, you can have your perceptual
information; what the heck.

Re: Pop PCT

About a week ago I posted a set of statements of what I (a lowly, non-
applied, scientific PCT type working at a comfortable distance from where
the rubber meets the road) took to be the basic insights of PCT with respect
to human nature. I was hoping that doing so would 1) start a discussion of
how PCT is applied in the "real world" and 2) how (and whether) these
insights are presented in the "Pop PCT" books. Thus far, there has been not
one peep from an applied PCTer. What's up?

Re: sci.cognitive

Speaking of no peeps, there has not been one reply or inquiry -- nothing in
response to my description of PCT in the sci.cognitive newsgroup.

Sometimes, being a PCTer is about as much fun as being a Maytag repair
man :-(.



<[Bill Leach 940407.20:07 EST(EDT)]

[Rick Marken (940407.1030)]

<"Dirty side down">

I read with great interest the postings by Dag (and wish that Ed would
get in here more too).

As I stated a couple of messages back, something to the effect of: "It
is really easy to make errors almost as soon as you leave the simple
single control loop."

You can add that it is also easy almost beyond belief to be
missunderstood even when you are right (assuming that you are).

I think that POP-PCT will always suffer from the terminology/reference
point problem.

PCT at the congitive level is a mix of perceptual signals, some derived
from level 1 perceptual functions and some derived from "memory".

In addition, there are probably multiple "programs" running at the same
time that are related.

Such concepts as being able to willfully set a reference for a perception
that is an "opinion" are tough to deal with from any view point including

I too hope to see the POP-PCT stuff discussed. It is likely important to
the future of PCT not too mention that "Dag's problems" are ones that I
personally relate to.