Reorganization, dependence

From Greg Williams (921012)

Bill Powers (921011.1500)

If you really received such a letter and actually believed it, would you
still say "Thank you for the information" and simply alter your behavior? Or
do you suppose there there might be a tad of affect attached to the

As I said in my previous post, I would probably be upset, but that doesn't
necessarily mean I'd be reorganizing. I suspect that affect (of whatever kind)
isn't always a sign of reorganization. If you disagree, I guess that's another
aspect of HPCT needing to be fleshed out with empirical data.

What starts and stops reorganization if critical error doesn't drive it?

For the sake of argument, I'll grant that reorganization starts when critical
errors get too big (relative to criteria of the organism at that time) and
stops when critical errors are reduced sufficiently (relative to criteria of
the organism at that time). Where do the criteria come from? They could be
inherited and unchanging, or they could been influenced also by the history of
the organism itself and/or environmental disturbances. Aside from that second
possibility, critical error involves two terms: a reference value for some
critical perception and (via subtraction in the PCT model) the perceptual
input currently extant; the latter explicitly depends on the independent
environment if control is not good (error not equal to zero), as is the case
when reorganization is occurring. So, except for solipsists running solely on
the "imagination connection," starting and stopping of reorganization depend
on BOTH internal AND external events.

Why would reorganization take place in response to "Fire" if the person
already had the goal of exiting upon hearing "Fire" and immediately did so?

It wouldn't, I think.

How can you be upset (experiencing critical error) without reorganizing?

I don't equate being upset and experiencing critical error. Given that: For
example, by not being sure you'll make it out of the theater while immediately
TRYING TO exit upon hearing "Fire," if you already had the goal of exiting
upon hearing "Fire" (and not needing to reorganize). For example, by not being
sure that you could control SUCCESSFULLY for getting your money back when
being confronted with the revelation of a con scam (which revelation you then
believe to be true) while beginning to try to get your money back, if you
already had the goal of getting money back which has been taken unlawfully
from you (and not needing to reorganize). I don't see reorganization as
necessary if the SPECIFIC disturbance to which I suddenly must attend
("offset" to maintain control) doesn't seem intractable, relative to my
current control structure. I do believe there can be anxiety WITHOUT
reorganization, simply because the success or failure of control isn't
always immediate or guaranteed, and it can be difficult to predict the outcome
in advance.

How can you control differently without acquiring a new control system or
modifying an old one?

"Differently" can be defined in terms of changes in the control structure. You
can only control for SOME perceptions with a particular control structure. To
control for OTHER perceptions ("controlling differently"), I agree that you
MUST "acquire a new control system or modify an old one."

How can new information lead to controlling something new without

That depends on what you mean by "controlling something new." If you end up
"controlling differently" as I just discussed above, then there must be
reorganization. But if the new information results in perceptions which are
able to be controlled by the current control structure, but which weren't
"actively" being controlled before the new information was perceived, then
there need be no reorganization. If I already can control for getting my money
back from a crook, when I learn that you're a crook, then I don't need to
reorganize, just "actively" (try to) control for getting my money back from
YOU, as a PARTICULAR crook -- no reorganization needed. I certainly might be
upset because of the (perceived) difficulties of controlling SUCCESSFULLY for
getting my money back. What a hassle!

How can a person use the fact that another person is reorganizing to control
for a particular behavior by the other person?

Ah, there's the rub: "particular." What I think A can do in some (make that
many) cases is arrange B's environment (disturb B) in ways so that B
reorganizes and so that the outcome of B's reorganization results in actions
by B which are in a class of actions as perceived by A which result in
perceptions A is controlling for. If B is ALREADY reorganizing (not due -- in
part -- to A's disturbances), I think that at least sometimes A's disturbances
during B's reorganization can result in B's post-reorganization actions
resulting in perceptions A is controlling for.

How this is possible is by A providing disturbances to B UNTIL B's actions are
in the class of actions which result in certain perceptions by A. Note that,
to the extent that the PATH of reorganization of B is unpredictable by A, A
cannot predict HOW LONG it will take to obtain a successful (to A) outcome of
reorganization (a good reason for "teaching" via sequentially arranged
"small"-distance reorganizations). But because A is CONTROLLING and hence
TRYING AGAIN when B reorganizes the "wrong" way (as A sees it), eventually, B
will act as A wants (not "exactly," but within a class of actions) -- unless,
of course, B doesn't want to play the game at all, in which case the
"teacher"/"counselor"/"parent"/"friend" A would be well-advised to wait until
B does want to play the game.

How can a person want another person to control his actions and at the same
time want the consequence that those actions are already controlling? Or are
you saying that there are actions which are not aimed at controlling


I don't understand the first question here. Please expand on it.

If A is partly determined by B, then to predict A from knowing B you must
also know the state of C and all other influences on A, known and unknown,
present and future.

Yes, to predict EXACTLY. How often in your daily life do you need to predict
ANYTHING exactly to see what you want? Never. You don't need to do that,
because you are a living control system, not a preprogrammed-output system.

How is your concept of "ceteris paribus" any different from the failed
methods of behavioral science?

If A controls for perceptions which depend on B's actions, that control can be
successfully achieved, in general, by numerous particular actions of B. A need
not predict or control for an EXACT output by B, only for perceiving AN action
of B which is A member of the class (defined by A) permitting control by A of
certain perceptions of A. A doesn't require PRECISE prediction of B's actions
to have successful control of preceptions depending on B's actions.

Where Skinner, in particular, failed was in not understanding that making
generative models of the underlying processes provides a means of explaining
the CONSTRAINTS on such control. But he wasn't interested in such
explanations, so TO HIM it wasn't a failure. If you asked him why THIS rat
(recently stuffed with Rat Chow) won't do the Skinner-box tricks that THAT rat
(starved for a while) will do, he would have said he didn't care -- but that
if you gave him both rats for a couple of days, they would both be doing even
MORE tricks than the starved one is doing now. That's "prediction and
control"! Skinner-nirvana.

How about reasoning it out? If A depends not only on B but on other variables
as well, and you do not know the states of all the other variables on which A
depends and cannot predict future states of all those other variables, how
can you predict anything about A?

You can't if all those other variables are making A fluctuate chaotically. But
if they DON'T (often the case -- and YOU define the "a lot"), then you can
predict future states of A ADEQUATELY FOR YOUR PURPOSES. Trust me, I was
trained as a mechanical engineer. How many bridges have you been over which
DIDN'T collapse under you? Their designers cannot predict exactly when they
WILL collapse (though they are pretty confident they will collapse before t =
infinity). And they are also pretty confident that they won't collapse soon,
as they define "soon." Maybe they're just fooling themselves. And maybe you're
just a minority of one on this issue. Nevertheless, as Steve Earle says (sort
of a secular Pascal's wager), "Just because you ain't paranoid don't mean
they ain't out to get you."

How can you determine that an apparent relationship is real?

With control, you don't need to know the causes of disturbances. You only need
the connection between your actions and your perceptions to be somewhat non-
chaotic. For reorganization, the setting of a problem with a particular CLASS
of solutions (i.e., ANY WAY you press that bar gets you food) is what makes it
sufficiently non-chaotic.

Best wishes,


[Martin Taylor 921012 15:05]
(Greg Williams 921012)

Good comments to Bill on partial prediction. A technical quibble, though.
You equate "chaotic" with what I would call "random." It is because the
world IS chaotic, not because it is not, that the engineer can predict that
the bridge will not fall "soon" but will fall at some indeterminate time
in the future.

Chaotic means, loosely, that short term prediction is normally pretty good,
but long-term prediction is impossible. My weather forecast for the next
millisecond is essentially perfect, for the next hour not bad, for the next
day better than chance (I can often decide reasonably that we will have
a picnic tomorrow), and for next week pretty useless. Professional
forecasters do better, but can't go beyond about a week. The weather is
a chaotic system.

In a random system, knowing the state at moment t1 gives you no information
about the state at any later moment t2, no matter how close t2 is to t1.

It is the chaotic nature of the world that permits PCT to work, and at the
same time allows short-term planning and habitual action patterns to be
useful. If the world were not chaotic, PCT would be unnecessary, since
pure planning could work. If it were random, PCT could not work. PCT
depends on the chaos of the world for its ability to stabilize perceptions.