Causal Control

[From Rick Marken (930914.1900)]

MIchael Fehlng (930914 3:31 PM PDT)--

Please recall my point (that appears to be lost among these wheels) is that
this _simplification_ has led some on this to a claim about PCT as a
non-causal theory.

That would be me, for one. In what way do you think that PCT is a causal
model?

That is evidently false.

Maybe it's evident to you. But it's not evident to me. To see why not,
read (if you want to, of course) my two papers in section 3 of
"Mind Readings".

I find
these attempts to drive a wedge between PCT and other theories regarding
causality are misleading and contentious. Such claims don't foster
understanding of the many, deep insights to be gained here.

What if the behavior of a negative feedback closed loop system happens to
be non-causal? Should we lie about it or conceal it?

We had a discussion of the causality issue earlier this year. It got to
the point where we were running models and posting raw data (columns of
numbers) to the net. The question being tested was whether the perceptual
signal is the cause of the outputs that oppose the disturbance to the
controlled variable. For example, in a tracking task, is the perceived
distance between target and cursor the cause of the handle movements that
keep the cursor near the target. The result of all this modeling and
debating was that those who believed in the perceptual cause of output
before the debate believed in it after the debate; and those who believed
in perceptual CONTROL (not causality) before the debate believed in it after
the debate. This is a VERY tough nut to crack -- and I think it is why
PCT people are considered so contentious (even by people in our "camp").
I believe it's not because we are particularly contentious (well, maybe
I am -- but Bill? Tom? naaaa). It's because the model itself is contentious;
it just doesn't work the way it "should".

I think there is only one way to dispell this image of contentiousness;
lie about the model. So, in the interests of agreeability, I hearby
proclaim that, in a control model, the perceptual input is the cause
of behavioral output. Control theory does not differ fundementally
from any other theory of mind or behavior; it just uses some different
words and emphasizes some interesting stuff. Other behavioral theories
can easily be incorporated into the PCT framework -- and PCT can become
a component of any other theory. They're all basically the same.

There, wasn't that nice.

Best

Rick (and yet it controls) Marken