[From Rick Marken (991210.1940)]
Bruce Abbott (991210.2010 EST)]
you shouldn't need a formal model to see that if the system in
question has another means of getting to the same end, one that
does not involve the variable being held to a fixed reference,
that the conflict between systems can easily be resolved.
In the spreadsheet exercise, three of the higher level systems
are not affected by the "committed" reference; these systems
can use other means (besides the reference for the "commited"
system) to acheive their ends. It's even possible, though unlikely,
that the "committed" reference will have no effect at all on the
performance of any of the higher level systems; this will happen
if the "committed" system is not used in any important way
as the means for any higher level system to achieve its end.
The point of the exercise was to show what _could_ happen (and,
indeed, is very likely to happen) if a child actually managed to
commititself to controlling a perception at some reference level
selected by some outside agent. Commiting to keep a perception
at some level may seem like a good idea to the agent but it may
end up being a very poor idea for the hierarchical control
system (child) who commits to it. Getting a child to commit
to control some perception -- for reasons other than those of
the child itself-- can wreck the child's ability to control (if
the child is a hierarchical input control system).
What's loose about it [the spreadsheet model]?
There are several skipped levels, for one thing. No
reorganization, for another. No memory, no imagination mode.
The only thing really HPCT about it is that three control
systems are stacked such that the higher levels control
their variables by manipulating the references of the
next-lower control systems.
And the perceptions at each level are of different _types_.
And there is an imagination mode. But what is in the model
certainly demonstrates my point, viz., if a person could
actually commit to controlling some perception at an arbitrarily
selected reference level it would almost certainly lead to
a deterioration in the person's ability to control perceptual
variables that he used to be able to control.
Bill says that he is satisfied that his analysis is technically
correct. That may be true. But if the situation is badly modeled,
it doesn't matter that the derivations from that model are
I think I have a way to stop our fight about whether we have
modeled any particular RTP practice correctly or not. I
suggest that we start by trying to answer this question:
What is it about the PCT model that led the developers of RTP
to decide on _any_ particular practice? For example, what
was it about the PCT model that led the developers of RTP to
decide that an important part of the program would be getting
a child to keep a commitment?
I'd like to see how applications people go from model to
practice (as I presume they did).