"standards" and bugs

[From Rick Marken (920522 08:40)]

The computer here will be down throughout the vacation and I
will be out of town until next friday. So if you have any
great thoughts about "standards" or other things that I should
go off and mull over, you'd better post them today before 15:00 PDT.

One thing that did occur to me that might be worth mentioning is
the relevence of this discussion of "standards" to one aspect of
the Beer bug discussion of long ago. For me, one of the most interesting
"revelations" of the Beer bug discussions was that rather complex output
systems can be involved in the control of rather "simple" sensory inputs.
For example, beer's bug had a rather complex, largely open loop gait
generation system that is really a means for moving the bug so that
it could influence (and control) some simple sensory inputs (like the
degree of bend in an antenna -- a unidimensional variable). What is
interesting to me about this is that an observer of the bug will see
some very complext behavior that could be described in all kinds of
complex ways (in terms, for example, of spatial variables that are
not even perceived by the bug). Yet, the actual "behavior" of the bug
(from a PCT point of view) is that it is keeping an intensity input
at a reference value.

The relevence of this to "standards"? I think when we look at social
behavior we are in a similar position to the observer of Beer's bug.
We see all kinds of interesting things happening -- but it is very
difficult to see the possibly very simple sensory variables that are
being controlled. PCT is an attempt to help us see beyond our
interpretations of behavior -- to what behavior is really about; control
of perception. And this requires a special kind of looking (based on
hypotheses about what variables might be controlled) and testing (to see
if disturbances to the variable are resisted). Just as we can read a
lot into the bug's behavior that is not relevent to what the bug is actually
doing (controlling) (for example, we see the bug "navigating" and "exploring"
in two-space, when, in fact, all it is "doing" is keeping the value of
one or two input variables at fixed reference values) so we may
also be reading a lot into human behavior that is not relevent to what
a person is controlling -- for example, we say a person has "bad manners"
or "poor standards" when they eat with their mouth open -- when, in fact,
they may just be controlling the amount of pain they feel because they
have a toothache.

It is hard to get past our inclination to see behavior as "output". We
assume that what we see is what the person is "doing". PCT suggests
that we must TRY to get over that inclination (if we want to understand
behavior) and take seriously the proposition that what we are seeing
(when we see people "behave") is the means by which people are keeping
their OWN perceptions matching their own references for these perceptions.
We, as observers, CANNOT see what another person is perceiving or
trying to perceive. We can only try to get an idea what a person might
be trying to perceive by doing the test for the controlled variable.

What seems to an observer as control of a complex principle (at the
"wrong" level with respect to the observer) may, in fact, be nothing
more than efforts to get from point A to point B in the context of
variable disturbances.

I've still got to work up a demo of this. Though Dag Forssell does an
excellent job of illustrating this point with his wonderful variations of
the "rubberband" demo; extroardiarily complex "behavior" seems to be
going on when people are doing nothing more than trying to perceive a
simple relationship between configurations -- "knot on dot".

Best regards




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