SDTEST3

[From Bill Powers (950308.0530 MST)]

Bruce Abbott (950307.1500 EST) --

Tried your new SDTEST3.PAS. Very interesting. Even more interesting when
you turn off (or ignore) the color changes. The intermittent
accumulation of scores can cause considerable confusion if the delay
between condition switches is comparable to the average delay between
scores. The reason is that you can switch to the other target, find that
nothing happens, switch back, find that nothing happens, and then start
switching at random to find the right target. When something finally
does happen, if the target changes again right away the "assignment of
credit" can be wrong.

Playing around with this experiment has led me to think some more about
the theoretical position taken in EAB.

ยทยทยท

------------------------------------
Let's look at an atheoretic description of the experimental situation
and the results.

Experimental conditions:

E1: If condition 1 holds, moving the cursor to the left target will
cause an increase in the score AND the cursor is red.

E2: If condition 2 holds, moving the cursor to the right target will
cause an increase in the score AND the cursor is green.

Results:

O1. If condition 1 holds the participant keeps the cursor mostly next to
the left target AND the cursor is red AND the score increases.

O2. If condition 2 holds the participant keeps the cursor mostly next to
the right target AND the cursor is green AND the score increases.

These statements of experimental conditions and observed results are
simply reports of what we can see. No theory is involved. There is no
mention of "perceptual signals", "reference signals", or organismic
"functions". The terms "reinforcement" and "stimulus" and "control" do
not occur. No explanation is offered. We are observing, not inferring.
-------------------------------------
The reason I went through that is to make it clear that when EAB types
speak of reinforcement and stimulus control, they are adding theoretical
constructs to the observations. When we build a control-system model of
the same behavior, we are also adding theoretical constructs to the
observations. But they are different constructs, or so I think.

I will try to construct a working model of this behavior.
----------------------------------------------------------------------
Best,

Bill P.