Tests and Rubberbands

[From Rick Marken (951111.1111)]

I said (to Bruce Abbott):

You described four things that you think are wrong with psychological
science. Not one has anything to do with the possibility that the system
under study might be controlling its perceptual input variables.

Bill Leach (951110.23:48 U.S. Eastern Time Zone) says:

Rick, it seems to me that 1 &/or 2 address this even if not as explicitely as
you might like to see

I don't see how points 1 and 2 address "the possibility that the system
under study might be controlling its perceptual input variables". Bruce's
point 1 was:

Students are not properly grounded in the the physical sciences,
mathematics, and the relevant engineering principles

This strikes me as irrelevant. I think many students of psychological
science are quite well grounded in the physical sciences and the relevant
engineering. Nevertheless, they continue doing psychological research
that ignores the possibility that the systems that they study might be
controlling perceptual variables. Bruce himself is quite well grounded in
physical sciences and the relevant engineering yet he continues to argue
that The Test for the controlled variable should not be central to the
approach to psychological science described in his research methods text.

So, while grounding in the physical sciences and relevant engineering
might be nice, it doesn't guarantee that researchers will seriously consider
the possibility that the system under study might be controlling its
perceptual input variables

Bruce's point 2 was:

Psychological researchers in every field have not yet recognized that the
key to understanding purposive behavior is the control system. And
because of Point 1, they are generally unprepared to construct control
system models or to conduct a proper analysis.

This point does get close; the control system is the key to understanding
purposeful behavior. But once you know how a control systems works
(control of perception) and you know that organisms _might_ be control
systems, then you know that you have start addressing the possibility
that the system under study might be controlling perceptual input
variables. You don't address this by constructing control systems models
(you don't know what to model until you do The Test) or by conducting the
"proper analysis" (you can't do the proper analysis -- one that determines the
variable(s) under control -- if data relevant to controlled variables is not
available; the relevant data is only likely to be available if you have best
testing for controlled variables). The ONLY way to address the possibility
that the system under study might be controlling perceptual input variables
is by testing to determine _whether_ the system (organism) is controlling its
inputs and, if so, _what_ inputs it is controlling. In other words, you have
to start testing for controlled variables -- the very approach to psychological
science that Bruce says he won't teach in his psychological methods text.

Gary Cziko (951111.1722 GMT)--

Re: The Lessons of the Rub

Thank you, Lord (er...Gary). Excellent post (the tithe is in the mail;-))

Best

Rick

<[Bill Leach 951111.21:09 U.S. Eastern Time Zone]

[Rick Marken (951111.1111)]

You waited until that time just so you could post with all of those
"ones", huh?

Ok, I see your point but still disagree primarily because I don't agree
that "many students ... are well grounded in ...". Hell, I don't think
that many students in the physical sciences are well grounded much less
those that are not. It is pretty bad when control systems engineers
don't even know what a control system actually is!

I also don't agree that Bruce asserts that THE TEST is not central to the
study of behaviour. THE TEST _is_ push-watch even if its' use (and
interpretation) is fundamentally different than "similar" use by others.

Your arguement with #2 is more relevant in my opinion. It is true that
experiments and data collection conducted done without an understanding
of the nature of control is likely to be useless. However, the
suggestion that behaviour is a control system phenomenon also suggests
that the construction of experiments would be oriented toward the study
of such systems.

-bill