[Martin Taylor 920325 17:00]
(Rick Marken 920324 18:00)
Methinks thou dost protest too much, mine Rick.
I have not observed any writer on this group who seemed to me to deny the
primacy of closed loop effects in the interpretation of behaviour. The
only question is whether ANY action is possibly open-loop at SOME level of
the hierarchy. I suppose there may be some minor disagreements about whether
there exist open loops that are closed by evolution rather than within the
living organism. Your rant is way, way, off the point. You take any
discussion of the possibility that some consistently observed relation between
a definable stimulus and a describable constellation of actions might be
open-loop as an assault on your faith. Heretics must be burned!
Speaking for myself, I think I have pretty completely absorbed and integrated
into my natural ways of thinking the concept that all behaviour is in the
end the control of perception. It makes so much of what we observe people
to do intelligible. But there are cases, such as target-shooting, where
the control is exercised over an outer loop that contains NECESSARILY open-loop
action elements. You can't correct the path of the shot bullet, so you had
better predict where it will go before you press the trigger. You can
certainly control the aiming point through the sights IF the sights are
reliably fixed (massively, as Bill puts it when describing possible conditions
for open-loop behaviour to be effective) to the gun. But only the next bullet's
strike can be affected by perception of the results of the last shot.
Open-loop, predictive actions are a part of this whole game. The fact that
all behaviour comes down to the control of perception doesn't falsify the
notion that there can be pretty complex open-loop constellations of actions.
Here are some test questions for those of you who think open loop
processes are important in some way.
What is your reason for even reading this newsgroup?
Could it be that the great unwashed might want to learn why they might want
to be baptized?
It's pretty obvious that you and Bill disagree as to whether there exist
ANY examples of open-loop constellations of action. The converted are
usually the most rigid of missionaries.
To put another cat in the neighbourhood of a pigeon roost, Bruce Nevin a few
days ago said something to the effect that PCT justified his failure to
learn about statistics in his youth. Rick and Bill (Bill less so) have
both inveighed against statistics. It is my belief that you CANNOT understand
PCT, and especially reorganization, without a reasonably intuitive
understanding of statistics. Any attempt to develop PCT deeply, beyond
simple tracking tasks, is going to founder on improper appreciation of the
statistical problems involved in the perceptual functions, if those problems
are simply defined out of existence. My modularity argument for reorganization
comes directly out of statistics, and I think most of the structural features
of the hierarchy will ultimately be derivable from the statistics of the
environment and the purposes that organisms can try to achieve.
Rick, do you want to issue another exclusionary Bull against statisticians?
Here's your chance.