Bridgman Article

[From Rick Marken (920505 13:00)]

Gary Cziko and Marcos Rodrigues suggest a CSG response to a paper posted
in PSYCHOLOQUY (a Refereed Electronic Journal) by Bruce Bridgeman called
"ON THE EVOLUTION OF CONSCIOUSNESS AND LANGUAGE".

I can see why; this paper says a lot of things that sound very consistent
with PCT. For example, the first sentence of the abstract says:
"Psychology can be based on plans, internally held images of achievement
that organize the stimulus-response links of traditional psychology".
Sounds like Bridgeman's "plan" is like a PCT "reference signal". Moreover,
Bridgeman seems to understand that "the dominant metatheory in psychology
has been the stimulus-response link" and that even "modern cognitive
psychology retains a similar orientation...". So Bridgmenan (a professor
of psychology at UC Santa Cruz) acknowledges what many critics of PCT
deny -- that scientific psychology is built on an input-output model of
behavior.

Unfortunately, after the excitement of the abstract and first couple of
paragraphs, it becomes clear that Bridgeman's model (to the extent that
one can understand it -- the descriptions are quite vague) is an "output
generation" model. A plan is not a reference for a particular level of
an perception, but a recipe for intended actions. In section 2.6
Bridgeman says "As a plan is executed, a single goal state or idea is
unpacked into a series of actions". BZZZZT. Wrong. As we (PCTers) know,
there is no way to "unpack" an internally specified goal into a particular
set of actions and have any hope that these actions will produce the
intended (goal) result with any consistency; actions produce their
results in a constantly changing environment so actions MUST VARY if
they are to produce the same goal result consistently.

Brigeman does make statements that suggest that, somewhere in the
back of his mind, he knows that his "plans" must be plans for
perceiving, not acting. In section 2.7 he says that plans must be
monitored -- which could mean that their sensory consequences are
important. But this is never stated clearly or quantitatively.
And in section 4.3 he says that motor commands are not commands for
specific muscle tensions (at least not the ones originating in the
cortex) but they are "an image of intended achievement" -- YES!!!
But he never explains the implications of this remarkable statement.
If the intended achievement is not a perception then how does this
image of intended achievement get transformed into the actual actions
that achieve it. The answer to this question is provided by HPCT --
the hierarchical model of perceptual control. I suspect that
Bridgeman doesn't even know that this problem (of turning intentions
into actions that produce consistent results in a disturbance-prone
environment) even exists for his model. Part of the problem is that he
has not tried to build a working model -- just a pleasant sounding
collection of words.

The basic point of his paper is completely unconvincing to me -- it
is simply the assertion that the word "consciousness" refers to the
process of carrying out these "plans". I don't see any relevant evidence
for this assertion at all. If a plan is really a perceptual control
system (giving Bridgeman the benefit of the doubt) then I disagree since
there are so many obvious examples of control taking place with no
consciousness whatsoever (in fact, we are unconscious of most of our
controlling -- such as controlling our balance).

So all in all I rate Bridgeman's paper a good example of the "close,
but no cigar" phenomenon.

Dag -- Great post on system concepts and standards (920504 -- I think).
Really excellent. I will try to give you some replies ASAP but I think
I agree with everything you said -- and said very well, too.

Regards

Rick

ยทยทยท

**************************************************************

Richard S. Marken USMail: 10459 Holman Ave
The Aerospace Corporation Los Angeles, CA 90024
Internet:marken@aerospace.aero.org
(310) 336-6214 (day)
(310) 474-0313 (evening)