NetNews Usenet Archive 1992 #31/NN_1992_31.iso/spool/bit/listserv/csgl/1996

Text File | 1992-12-28 | 5.4 KB | 115 lines

Comments: Gated by [e-mail]
Path: sparky!uunet!paladin.american.edu!auvm!COURIER4.AERO.ORG!MARKEN
X-Delivery-Notice:  SMTP MAIL FROM does not correspond to sender.
Posted-Date: Mon, 28 Dec 92 09:43:38 -0800
X400-Trace: US**AEROSPACE; arrival Mon, 28 Dec 92 09:43:38 -0800 action Relayed
P1-Message-Id: US**AEROSPACE; 921228174338
Ua-Content-Id: CSI NC V2.1b
Message-ID: <0002EA36.MAI*[e-mail]>
Newsgroups: bit.listserv.csg-l
Date:         Mon, 28 Dec 1992 09:43:38 -0800
Sender:       "Control Systems Group Network (CSGnet)" <[e-mail]>
From:         [e-mail]
Subject: Re: PCT Fanatics
Lines: 99

[From Rick Marken (921228.0900)]

Oded Maler (921226) --

>Your main problem
>is not being able to go up one (or several) levels above your
>background as an ex-"scientific psychologist" and realize that
>some people are simply asking different questions or have different
>scientific goals than "building a predictive theory of human behavior".

I don't believe that I have ever taken issue with people whose
scientific goals are not obviously relevant to PCT. When have
I done this?

>So it's no use attacking this or that person/theory immediately
>each time you detect the he/she/it has not realized what behavior
>is, etc. People who are doing, say, mathematics of non-linear dynamics
>are interested in general properties of some systems obeying certain
>rules. Period.

Perhaps I have not been clear; I have no beef with non-linear dynamics
as an area of study per se (nor with any of the other examples of what
I have called "trendy" theories). I just assumed it would be understood
(from the nature of this list and from the context of the posts) that I
was taking issue with the theory or formalism or whatever when it
was being proposed as a MODEL of some aspect of purposeful behavior.

>Although some others may try to apply this math to
>psychology by using the wrong (i.e., non-PCT) model of behavior, it
>does not mean that some fundamental truths about such systems are
>not relevant and will not be needed when more complex PCT models will
>be built.

OF COURSE. But this list that is about PCT and purposeful behavior
(control). There is no rule against posting interesting findings about the
behavior of non-linear dynamical systems or whatever. But I can't imagine
that there will be too many people who will be that interested in those
findings per se unless their relevence to modelling and understanding
purposful behavior is fairly clear NOW.

>If we take your favorite mathematical formalism, linear algebra,
>does the fact that arithmetical operations are also used in analyzing
>behaviorist models, make them uselss in modeling?

Of course not -- and the tools themselves are important, of course. I
don't believe I have ever criticized the tools themselves -- why would
I? I only know about most of these tools because of the efforts of various
people to use them as models of purposful behavior; in that case,
the people who use these tools have goals that are the same as mine
and I think they are trying to achieve that goal in the wrong way; so
I criticize the tool as a MODEL OF BEHAVIOR, not as a tool per se.
I don't criticize statistics per se, for example, just because I think
it is generally irrelevant to understanding the purposeful behavior of
individual organisms.

>The same goes with your outrage toward roboticist until you
>realized that they have other (orthogonal..) goals, namely to
>build toys and not to analyze existing living systems.

I don't know about "outrage"; when roboticists try to build purposful
systems using the wrong approach, then I think it's OK to say what's
wrong. When the goals of the roboticist (or anyone else) are truly
orthogonal to those of the PCTer then there is no disagreement. But
sometimes "orthogonality" is in the eye of the beholder -- the cry of
"orthogonality" can be used to stifle legitimate dialog. For example,
you say:

>The same
>goes with Braitenberg, whose main occupation is experimental
>neurobiology and his little "vehicles" book was just written for
>fun and speculation (and yet it has some interesting ideas,
>including the imagination-loop idea used in planning) and not
>as a serious suggestion of an all-embracing theory of behavior.

Does this mean that it is not legitimate to show that Braitenberg
has ignored the fact that his vehicles actually operate by controlling
sensory variables? If Randall Beer said that he was just building bugs
for fun should we have just assumed that he was using the right
approach? If roboticists are just trying to build machines that
produce particular (purposeful) results using inverse kinematics,
should we say nothing just because this is what they want to do -- ie.
they want to figure out faster, more efficient algorithms for doing
inverse kinematics -- even though we know that output generation
is, in itself, the wrong approach to producing purposeful results?

If a person just cares about the tool per se (info theory, statistics,
detection
theory, non-linear dynamics, transformational grammars, etc) then I have
no beef -- never have, never will. When they start applying the tool as
an explanation of some aspect of purposeful behavior, then I feel like it's
OK to criticize the use of that tool from a PCT perspective.

The life sciences have used the plea of "orthogonality" for years as
a means of avoiding a confrontation with PCT; they just say that purpose
is orthogonal to what they are about. Pardon me if I disagree.

Best

Rick