Modeling SDTEST3 Performance

[From Bruce Abbott (950313.1325 EST)]

Rick Marken (950312.1630)

Bruce Abbott (950312.1410 EST)

I'm 49 too . This leads me to suspect it's not age, it's maturity. Yes,
response speed must be inversely related to maturity. That would certainly
explain why you and Bill P. are so slow, why Mary is pretty slow and why I
am so fast. Now we need to know not only how old Sam is but, more
importantly, whether he is anywhere near as immature as I am;-)

I don't know about maturity, but I think I should probably avoid playing
racketball with you. The ball will be hitting me in the face before I
realize it's coming at me. Maybe the operative variable is the ages of our
kids; the young ones help to keep your low-level systems fine-tuned for fast

I envision a component of the model that selects left-target when the
cursor is green and right-target when the cursor is red.

This is a nice S-R version of the model: the perceived color of
the cursor selects the reference position of the cursor.

Sorry, doc, I'm back-sliding again. Can you direct me to the nearest
chapter of AA (Animists Anonymous)? I really meant nothing more here than
that there is a link between perceived cursor color and perceived active
target without committing to a specific mechanism that would accomplish the
linkage. In fact, I was invisioning your control system suggestion as one
possibility while keeping other possibilities open. And I agree that the
Test is the way to find out.

This suggests that the initial mechanism might be feedback regulated

If by "feedback" you mean the perceptual consequences of action then,
in a feedback loop, feedback is regulated (controlled) -- it does not
regulate. If, however, you are using "feedback" to refer to what it
is -- the effect of a variable on itself via a closed causal loop -- then
I suppose it is true that feedback "regulates"; but it doesn't regulate
a mechanism -- it regulates one particular variable in the feedback
loop: the perceptual variable.

The phrase "feedback regulated" is a red flag for me because it is the way
conventional psychologists talk about control. They use this terms because
it suggests that perception (input) guides (regulates) output (repsonses).
It would be nice for them if control worked this way -- but, infortunately,
it doesn't.

Just slipping into old habits again. (My intro to control theory was Ashby,
not Powers, where I acquired terms like that one.) In a forum where
"control" refers specifically and only to closed-loop, negative feedback
control, the term "feedback regulated" becomes superfluous as a qualifier
indicating the type of control. I was trying to distinguish closed-loop and
open-loop systems; by "feedback regulated" I meant only the usual control
system we always talk about.

You mean practice would turn the task into an open loop, S-R task?
Why would practice do that -- except to make some conventional
psychologists happy;-)

It might be highly adaptive. Building the linkage would require control:
perceiving the relationship between color and selected target and correcting
errors of selection. Sufficient repetition might establish a direct
associative linkage; perhaps this is one way that some perceptual functions
are built up. An open-loop perceptual "translation" from color to selected
target would be very efficient of resources and speedier. Brief, infrequent
application of the Test might show that the linkage is not at this later
stage of practice actively controlled. However, application of the Test
would create a high-level error (in points accumulation) and trigger
reactivation of the supervisory control system if done too often, which
would lead one to conclude that the perception was under control the whole
time, so one would have to be careful to avoid this problem. Even so, my
proposal is testable, and I wouldn't rule it out a priori.

It will be great when you turn these guesses about how control
works into actual program code. Then we'll be able to see if we
really need PCT to explain "stimulus control" or whether the EAB
bunch had it right all along.

I'm afraid I don't see it in those black-and-white terms. Perhaps there are
elements of both views that are correct--I'm willing to keep an open mind to
that possibility. As for code, once I've developed a sensible model or
models to test, the code will follow. (;->