Confident Proclamations

[From Bruce Abbott (950705.1910 EST)]

Rick (a good guy) Marken (950705.0820) --

Bill Leach (950704.04:21)

What I think has to occur is that a sufficient precise symbolic EAB
system has to be created that exposes the existance of: "Albus type
models" (magic occurs here), use of "standard" terms that can not
possibly have the standard meaning when examined closely, use of
identities as causal explainations, hidden logical absurdities, and the
like.

Bruce (acting for the EAB side) has confidently proclaimed that EAB can
account for data that seems (to me, att least) to be completely inconsistent
with a reinforcent model of behavior (ratio data, random consequences,
changed reference state).

First, I think Bill is right that putting EAB statements in quatitiative
form will be extremely helpful in identifying holes and inconsistencies. As
to my "confident predictions," I believe I've only claimed that I "believe"
that a model based on reinforcement theory could be constructed that would
account for the ratio data. This is only my intuition based on my knowledge
of reinforcement theory; I could be wrong. As should be apparent by now

If he knows this much about the EAB and PCT models,
then he must also know what kind of data WOULD distinguish between the two
models. Yet Bruce has never proposed a test that would distinguish between
these models.

Actually, I don't know. If I did, I would propose such a test. All I can
say is that the tests you've proposed don't seem to do it.

Bruce has done nothing but defend the EAB models even though he
knows little about how these models compare to PCT (I say the latter because
Bruce now wants to get involved in a detailed analysis of the relationship
between EAB and PCT models of behavior to see how the two models really
compare).

Actually I've been more enthusiastic in defending EAB science than EAB
theory. In my view, researchers and theorists by and large have been doing
good work within the framework they inherited from those who came before
them, but they've been severely handicapped by a faulty set of starting
assumptions. What's going on now is that Bill Powers and I are working on
the project that will (soon, I hope) produce data on rat behavior on a
number of simple schedules of reinforcement. The ongoing theoretical
discussion concerning PCT and reinforcement models will help to develop
models whose implications can be tested empirically when we collect the
data. There may be conflicting predictions of the PCT and reinforcement
models which we will want to examine; if so we should tease them out prior
to collecting the data so that they can be allowed for in the design.

What totally mystifies me is why you object to this approach.

What are you afraid of?

Regards,

Bruce

ยทยทยท

from my descriptions of the theory, it's a pretty complex affair.