Why PCT? (was Re: Flies That Do Calculus With Their Wings)

[From Rick Marken (2014.03.26.1035)]

Richard Pfau (2014.03.24 17:05 EST)

RP: Bruce and Rick,

I hope that one or both of you contact the authors and inform them about the relevance of PCT to their scientific paper

RM: Both Bruce and I said that it was probably not worth it to contact
these authors. And I think we both based that conclusion on the fact
that the authors clearly understand control theory and PCT is just
control theory. So there is really nothing to contact them about.

But this answer doesn't really satisfy me because I obviously do think
PCT differs in important ways from other applications of control
theory in the behavioral sciences. And there are many people in the
behavioral sciences (like those who did this Fruit Fly Pitch Control
study) who do use control theory as the model of the behavior they
study. Control theory is very big in the study in the study of manual
control, for example. There are many textbooks on the application on
control theory in the behavioral sciences. For example, there is
"Control theory for humans: Quantitative approaches to modeling
performance" by R. Jagacinski and J. Flach (NJ: Erlbaum, 2002). The
authors of this book, like those of the Fruit Fly Pitch Control study,
are behavioral scientists who know control theory, and they know it in
great mathematical detail -- Laplace transforms and all.

So there are all these behavioral scientists who understand control
theory quite well but they have no truck with PCT. Clearly, even
though PCT is control theory, these control theorists see that there
is something about PCT that they don't like. So what's up here? Why do
control theorists seem to reject PCT?

I think one way to get at an answer to this is to see why those of you
who are behavioral scientists and fans of PCT have chosen PCT over the
version of control theory that is favored by most behavioral
scientists. In other words, why PCT?

Best

Rick

ยทยทยท

--
Richard S. Marken PhD
www.mindreadings.com

It is difficult to get a man to understand something, when his salary
depends upon his not understanding it. -- Upton Sinclair

[Martin Taylor 2014.03.26.14.18]

[From Rick Marken (2014.03.26.1035)]

Richard Pfau (2014.03.24 17:05 EST)

RP: Bruce and Rick,

I hope that one or both of you contact the authors and inform them about the relevance of PCT to their scientific paper

RM: Both Bruce and I said that it was probably not worth it to contact
these authors. And I think we both based that conclusion on the fact
that the authors clearly understand control theory and PCT is just
control theory. So there is really nothing to contact them about.

... So what's up here? Why do
control theorists seem to reject PCT?

I think one way to get at an answer to this is to see why those of you
who are behavioral scientists and fans of PCT have chosen PCT over the
version of control theory that is favored by most behavioral
scientists. In other words, why PCT?

Best

Rick

Your question is quite difficult for me, as a behavioural scientist who contemplated doing post-graduate work in control theory. I can't really answer it, but I'll give you a kind of "this could be the reason" post-hoc excuse. What happened at the time might have been quite different, but this is how I remember it.

Starting around 1980, we had been developing a multi-level interface theory to allow us to program an interface that allowed the user to combine voice, keyboard, and pointing gestures in a single "sentence" when working with a military tactical display. This developed into an early version of the Layered Protocols Theory of communication. But the key difference between that version and the version that led me to PCT was that we/I initially saw the problem from outside, as a programming problem. We/I acted as an external observer or analyst, looking at both sides of the interactions at once and analyzing what was going on as an abstract physical process -- which is a perfectly legitimate way to look at it.

The breakthrough happened when we began to look at the interactions from the point of view of each actor separately. Instead of imagining a flow of information around multiple levels of loops that incorporated both partners at each level, we considered what each could perceive about the other as a consequence of the other's actions -- verbal or otherwise. That shift of viewpoint made the whole theory much easier to work with. It allowed for the development of the "General Protocol Grammar" which depends on considering the partners' viewpoints separately, and when I learned about the existence of PCT (from someone on the Systems Analysis mailing list), it very quickly became obvious that the LPT was simply a special case of PCT in which two multi-level control systems interacted in a common environment.

I wonder if the issue with getting control theorists to _feel_ PCT is a similar problem of viewpoint. When you analyse the flow of signals around a negative feedback loop, the mathematics is exactly the same as it is in a PCT analysis, so why change? The difference is in the ability to take the controller's viewpoint, the observer's viewpoint, and the analyst's viewpoint depending on what matters at any one moment. If you have only the analyst's viewpoint available to you, PCT has no advantage, because nothing changes in your understanding of the loop. If you say "the controlled variable is the one compared with the reference value", that's self-evident to any control theorist. What changes your understanding is if you say "This system can only _command_ its output; it doesn't sense whether its command is obeyed, because all it senses is the input that creates the controlled variable. This system cannot _control_ its output at all."

I don't know whether this is the source of the problem, but for me, the addition of the actor-centric viewpoint was the important shift. I didn't have to accept PCT. All I had to do was to recognize that it was a greater abstraction of LPT that put communication into the context of everything a communicating organism does. Others using control theory to analyze biological action may not have the same advantage.

We spent a lot of time in my early years on CSGnet talking about viewpoint shifts and their uses. That hasn't happened in the last decade or more. Maybe we all take it too much for granted.

Anyway Them's my thunks.

(Interesting -- the spell checker didn't complain about "thunks").

Martin

[From Adam Matic (2014.03.28 1030 cet)]

Rick Marken (2014.03.26.1035)
So there are all these behavioral scientists who understand control
theory quite well but they have no truck with PCT. Clearly, even
though PCT is control theory, these control theorists see that there
is something about PCT that they don't like. So what's up here? Why do
control theorists seem to reject PCT?

I think one way to get at an answer to this is to see why those of you
who are behavioral scientists and fans of PCT have chosen PCT over the
version of control theory that is favored by most behavioral
scientists. In other words, why PCT?

AM:
I really have no idea how can anyone reject a demonstration of a working multilevel system. I was captivated. I talked to a few engineers, and the reaction was 'meh, control of perception is a play on words and negative feedback is known for centuries, nothing new about this PCT'.
On why PCT and not other versions of control theory - they are just a bunch of math symbol crunching. Here is Carver Mead, about old-time engineering:
"Engineering wasn't something you studied and learned and memorized and knew where to look up. Engineering was understanding things all the way to the bottom, no matter what field they were called, and being able to use that to build stuff and make it work."

To me, PCT provides a road map for understanding control systems that no other theory offers.
Adam