[Fred Nickols (971204.0610)]
Mary Powers (971202)
To Hank Folson and Fred Nickols:
Hank, your posts on complexity not being the real issue and on the
difference between thinking of organisms as being control systems vs. merely
including them were very nicely put.However, I think both you and Fred have derailed in your recent discussion.
FN: 1) When you say behaviors are "chosen, is that chosen as selected from a
repertoire of possible behaviors, presumably for the purpose of conscious
execution with some outcome in mind?HF: That sounds reasonable for recurring situations where we can remember
the general control outputs that worked before and use one of them as the
starting point.MP: One of the principles of PCT is that we do not sense outputs - the
cascade of higher level outputs setting lower level reference levels down
through the hierarchy. The only outputs we sense are those we perceive.
What we remember are perceptions. We may remember perceiving a sequence of
movements that we want to reproduce, or the perceived amount of effort a
particular act required, but simply reproducing what we did before will
NEVER work -
CANNOT work, because initiating a movement is unlikely to start from exactly
the same position as before, and therefore requires different effort, and
doing it with one's muscle fibers in exactly the same condition and
configuration of fatigue and recovery is utterly, totally, unlikely. This
is the eye of the needle, so to speak, that all theories of behavior have to
go through - have to explain - and that most ignore. It is the core of the
concept that behavior must be the control of perception. It is not behavior
that is chosen, it is perception.
If I choose to raise my arm, and do so, I perceive my arm moving through the
air and I perceive it at whatever position I accept or have set as "raised."
I might also have some proprioreceptive sensations in the muscles and joints
that are associated with moving my arm (although in my case I suspect these
are mostly aches and pains associated with advancing years). It seems to me
that I can envision a path along which I want my arm to move and move it
along that path (plus or minus some variance). As I raise my arm, I aware
of its changed position in space and can choose to move it pretty much as I
see fit -- within whatever physical limits apply. It seems to me that I can
do that again -- and I will readily agree that it is highly unlikely that I
can do it in precisely the same way I did it before. I can reproduce in
general the act of raising my arm but, as Mary says, I cannot reproduce
exactly a previous act of raising my arm.
So, if I say that behavior refers to whatever internal activity is moving my
arm and that perception refers to the perceived movement and position of my
arm, then what Mary says above is true; namely, that "It is not behavior
that is chosen, it is perception." No problem in technical discussions, but
most people I know will have difficulty not viewing my arm moving through
space as behavior on my part.
FN: Or does chosen refer to some more basic, less conscious act or process?
HF: That sounds reasonable too. I do not know how we make our initial
decision to try a specific output to attempt to return a variable to its
reference level...MP: This is really a ninth level strategy-planning matter. At this level one
can imagine (perceive) acting in various ways and imagine the perceived
consequences. Setting reference signals for the lower levels and ultimately
the muscles that carry out the strategy chosen all involve unconscious
processes, and one is not likely to be conscious either of the perceptions
at lower levels involved in bringing the chosen strategy to the desired state.
No problem; see arm-raising example above.
FN: Are behaviors themselves perceived in light of some reference signal and
varied so as to be consistent with that reference signal?HF: Humans for sure have a complex enough hierarchy to do it this way. I
doubt that E. coli could do it this way. I would think this approach would
be useful, and the most efficient option as a starting point for outputs
only for routine repetitive actions.FN: Or do we simply vary our behavior so as to reduce the discrepancy
between some non-behavioral variable and its reference condition?MP: For the most part, it's the _consequences_ of behavior that one has a
perception of and a reference signal for, and the behavioral means is
unimportant. Sometimes, however, the desired consequence is the behavior
itself, as perceived and as wanted - performing a particular sequence of
steps in a dance, or doing a swimming kick properly, etc., etc. But what
you control is still the perceived consequence of behavior, not the behavior
itself, about which you know nothing except what you can perceive.
Mercy, mercy. My guess is that Bruce Abbott is going to have a whale of a
good time with the first sentence in the paragraph above. Anyway, I agree
with it subject to a stipulation that we often find ourselves in situations
where what we want is contingent upon some actions (not behavior) on our
part and so our behavior is limited to that which supports the range of
actions necessary to get what we want (assuming we want it badly enough to
try and get it). Whew! Hmm. I think I've just found a useful distinction
in my own thinking; namely, the one between behavior and actions. In the
arm-raising example above, the action is raising my arm, the behavior is
muscular and skeletal activity (I can't spell the combined word), plus some
other stuff about which I know absolutely nothing.
A repetitive action is a controlled sequence of events: hammering, beating
a drum. The relevant perception is controlled - doing "the same thing" over
and over, plus, in the case of the hammer, stopping when the nail is in, or,
in the case of the drum, maintaining a steady, even beat. However steady
the beat, you still aren't "doing the same thing" each time - different
neurons are firing, different muscle fibers are involved, for each stroke.
This doesn;t mean the beat isn't really steady - it is, as steady as you can
perceive it. You are accomplishing consistent ends by variable means.
Agreed and I don't recall that my behaviorist friends have ever adequately
accounted for successful hammering under varying conditions. I've always
thought that it was "learned" and used to use it and similar examples to
draw a distinction between natural reinforcement (i.e., successfully driving
the nail under varying conditions) and contrived reinforcement (e.g.,
dropping a pellet of money into the successful carpenter's pants pocket). 
Thanks, Mary, for a useful post.
Regards,
Fred Nickols
nickols@worldnet.att.net