I too palpably felt an epistemological fog settling in as I read
/*****************************************************************************
* *
* But it should all be understood in the context that the organism *
* knows nothing about perceptions. It knows only about error and *
* controls its perceptions because over time the control of those *
* perceptions has resulted in a less error-ridden state of being. *
* Sure we experience these perceptions, but the organism does not *
* Know about them, nor does it really "care" about them--it cares *
* about error. *
*****************************************************************************/
But it is useful to present new ideas in different ways and from
different angles, as Bill perhaps more than any has demonstrated
again and again in his writing here. Not merely useful, even
necessary. To "get it" subjectively, to really connect the
subjective here/now me this is it experience that I have (you
have, the person you are talking to has) with PCT it is
perception and not error that you "know about" and "care about".
And until that occurs and pointing out something to someone
becomes an even greater marvel, in the act of pointing, than the
thing pointed at <oops! recursion alert! > some basic
revaluation of system concepts has not yet taken place, some
revelation, conversion, recognition, assimilation, integration,
call it what you will. But on the way to that epiphany can be
many steps, and for some people introducing the signals in the
ECS model might be helped by saying "in an important sense we
could say that the organism controls error, though it sure feels
to us as though it is perceptions that matter." As to what
matters most, an argument could be made that it is the reference
perception that is the focus (of attention? As Poo would say,
`Oh help!'), and that real-time perceptual input becomes apparent
to us only by way of error. I'm not sure how well that would
stand up (more epistemological fog), but it seems at least as
plausible.
Bruce Nevin [921204.0910] -- nicely put! My declaration of defuddlement
was not intended as a criticism of people who are posting about "control
of error." I often -- very often -- have the same sensation while
writing about PCT -- my own words defy my understanding.
It is certainly true that a person can focus on the fact that control systems
reduce error to or near zero, then conclude that the systems control
error. After all, the numbers are there to see, in any simulation. And
Bill's demos (and Rick's spreadsheet) lay those nimbers out for anyone
to see, in real time. (Oops. I perceive an error, but I can't get back
to it on this editor! But the error I see -- nimber -- is not an error
in the environment. Out there, nimber and number are just stuff on a
screen -- not even that, if you press the issue. There is error only
if and when I decide I should see something otherthan what I do see. Oops.
There it is again.)
Error does not and logically cannot exist objectively or independently
(unlike, perhaps, some perceptions, but that's a different argument).
Error always exists REALTIVE to a given STANDARD: as a DEPARTURE from
some asserted, assumed, or given "correct" state. And what is that
state? The reference level. So while reference levels, as imagined
perceptions, can exist without errors, errors must have RLs to exist;
thus RLs are logically prior to errors.
Put another way, an error is a FUNCTION of two arguments: the imagined
RL, and the "real world" perception. An epistemology based on error
will beg the question as to the nature of the RL and the perception.
Thus in PCT (Powers, not Perceptual, Control Theory), it is
imagination and perception which must be primitive.
No "epiphany" or "leap of faith" is necessary to reach this
understanding. It follows simply from clear thought. Bill's knot shift
helps, too.
Cliff Joslyn, Cybernetician at Large, 327 Spring St #2 Portland ME 04102 USA
Systems Science, SUNY Binghamton NASA Goddard Space Flight Center
cjoslyn@bingsuns.cc.binghamton.edu joslyn@kong.gsfc.nasa.gov
Thus in PCT (Powers, not Perceptual, Control Theory), it is
imagination and perception which must be primitive.
Just to clarify where I stand: I do not disagree with anything you
say. The issue did not pertain to whether errors or perceptions are
"more primitive" than the other. The question was whether we intentionally
try to control our errors (and thus control perceptions -- as would
necessarily follow from the relationship of errors, perceptions, and
references), or we purposively control our perceptions. In other words,
what is it that the control system really cares about: the errors (i.e. the
difference between the reference and the perception) or the perceptions
themselves. Would a perception be controlled without a reference?
Anyway, I'll drop the subject as it not essential to the model. Just
wanted to let you know that as you presumed, I surely didn't need a "leap
of faith" to understand your argument.
Thanks, everyone, for the responses to my "purposive controlling" dillema.
> some basic