[From Rick Marken (950703.1140)]
Bruce Abbott (950703.1145 EST) --
Difficulties in dealing adequately with such constraints on a theoretical
level have led to considerable energies being expended to identify
alternative models that might handle the problem (in direct contradiction to
Rick Marken's assertion that reinforcement theorists simply "ignore" such
difficulties).
I was asserting that one particular "acting" reinforcement theorist was
ignoring the results of the random consequences demo where we showed that
people would still press the bar appropriately to keep the cursor on the
target even though the consequences of presses were random.
You had shown that a non-random consequence --change in direction after a
press-- could account for non-random pressing. Then Bill and I made the
change in direction random and showed that subjects can still produce the non-
random result (cursor on target). You have still not explained this result:
how does reinforcement theory explain the fact that random consequences
select non-random behavior.
You want a quantitative model of ratio performance?
Not me. I want ONE quantitative prediction of reinforcement theory that
DIFFERS from a quantitative prediction of PCT.
Me:
Bruce Abbott has been developing a novel approach to "nothing but"-ism that
I would call "nothing but, but better";
Bruce Abbott (950703.1230 EST) --
What I am astounded by is that you would say that, Rick.
I assume that you have drawn this conclusion because I have indicated that
certain parallels exist between the two views.
No. I have drawn this conclusion because we have been unable to point to any
data that seems, to you, to be inconsistent with reinforcement theory; the
ratio data, the random consequences data, the change in steady state cursor
position data, etc. So it seems to me that you see reinforcement theory, in
one form or another, as being an adequate model of behavior.
When I point to data that seems obviously inconsistent with the notion of
selection by consequences you are quick to point out ways it could be handled
by reinforcement theory. You have never said anything like "Yes, that looks
like a good way to show that consequences don't select responses. But a
reinforcement theorist might say 'such and such' so here is what you could do
to make the demonstration more convincing".
All I get from you are reasons why reinforcement theory can account for data
that seem to clearly reject reinforcement theory and support PCT. So it
seems like you believe that there is no way to empirically distinguish
reinforcement theory from PCT. That is why I have concluded that you think of
PCT as "nothing but" reinforcement theory.
Apparently, you believe that reinforcement theory and PCT can handle all the
same data. So I presume you believe that the only way to distinguish the
two theories is to see which seems simpler or more accurate. I think you
believe that PCT is simpler and possibly more accurate. In other words, PCT
is "nothing but" reinforcement theory (it handles all the same data), but
better (it does it more simply and accurately).
Same holds for your conclusion about my position: absurd.
It would be very simple for you to show that my conclusion about what you
believe is, indeed, absurd. Just present a study that will distinguish
reinforcement theory from PCT. You keep avoiding this request; I assume this
is because my request is a disturbance to what you are controlling for (the
belief that PCT is "nothing but" reinforcement theory, but better).
So please prove that my conclusion about your position is, indeed, absurd.
You can do this by giving me what I ask for above:
ONE quantitative prediction of reinforcement theory that DIFFERS from a
quantitative prediction of PCT.
Thanks
Rick