[from Jeff Vancouver 980219.0900 EST]
So many posts, so little time.
Apparently, I have launched a firestorm. This makes it difficult to bow
out gracefully. Indeed, I feel like I have made matters worse. So let me
proceed head long inserting the other foot (maybe I will reorganize the
unit that is driving this OUTPUT after this round).
First, what I meant:
[From Bruce Abbott (980218.2025 EST)]
Bruce A. to Bruce G.
Bruce, I take Jeff to be saying o = f(e) where o is the output of a
control
system, and e is the error signal. You seem to be denying this statement.
No, he is saying that output is not behavior. Producing output per se is
not
behavior. Rather, behavior is the control of perception.
Since this is what I am saying, I feel justified in saying that Bruce A. is
correct and Bruce G. is incorrect. Clearly, one of my mistakes is in not
using the equation o = f(e). Perhaps another mistake was in not using the
word output instead of behavior. But, I am not sure that was a mistake on
my part. Please read on.
>That depends on what definition of behavior you are using.
tenatively agreed.
On many
>occasions, Bill P. has equated behavior with output.
i. kurtzer
Bill is not PCT.
Bruce G.'s primary point seems to be that one should be precise in their
use of language. This is undoubtedly true. Yet, at the same time, I am
saying that specific words apparently have specific meaning to specific
individuals such that it is difficult to communicate because we do not have
shared meanings. In fact, one of my main points is that because of this
reality of differences in meaning I think it is useful to lighten up
somewhat in our posts. Now I am NOT saying that we should allow for
ambiguity and loose talk. What I am saying is that we should try to be
more understanding of the ambiguity of the words used. For example, my
"mistake" of using the word "behavior" instead of "output" is a good
example. What did I mean by behavior? What does Rick mean? What does
Bill mean? What does PCT mean (since Bill and PCT are not one, i.)?
I meant the output of a system. I believe this is one of the uses for the
word behavior that Bill has meant. To substantiate that belief let me
quote Bill from B:CP. "The purpose of any given behavior is to prevent
controlled percpetions from changing away from the reference condition."
Now if Bill is not PCT, what is? B:CP? Rick? I do not like Rick's
definition that behavior is the control of perception. I am much more
comfortable with behavior is the result of _and_ helps one control
perception. So let me try to be precise. There are actually three terms
of relevance here: behavior, action, and output. This is what I think
these things mean to PCT, starting from the last. Output is the signals
eminating from a function. Usually, the term is reserved, but not
necessarily, for the output from the output function. However, it has been
said, for example, that error (whether it is zero or more) is the output of
the comparator function. But even if we limit it to the output of the
output function, output is only observable when it is from the intensity
level units. Hence, I am not inclined to use the word output when refering
to observable actions on the part of the system.
Actions is generally the word PCT uses to talk about the outputs of the
intensity functions. They are what influence the environmental variable
(along with disturbances). They occur on and in the environment, as
opposed to outputs which only occur in the environment in the special case
of the outputs of the intensity unit output functions.
Behavior, is seems, has two meanings in PCT. The meaning that I prefer,
but Rick does not, is the observation of actions. Behavior is the word for
the actions the scientist (or whomever) observes. It is the observations
of Mary regarding the actions of posters like me that prompted this whole
thread. That is why I used the term behavior.
Rick, and sometimes Bill and others, take behavior to mean the entire
operation of the control system. Now, if one thinks about it, it may be
that these two meanings are not different. If the observer is Rick or Bill
or Mary, or another PCTer, their understanding of the actions is as a
consequence/result/sideeffect (what word?) of the closed loop of control
systems where the key issue is the control of perceptions. Hence, behavior
is the control of perception to a PCTer. However, to apply one's
understanding of the nature of behavior (even if we agree about it) to the
actions strikes me as adding a layer to the concept of behavior.
Now my guess is that no one will completely agree with everything I have
said about these terms. I expect no less. As i. Kurtzer points out, Bill
is not PCT. So what does that leave us. There is no definitive dictionary
of PCT terms. Each has there own meanings (if highly overlapping) which
are subject to change. What I am looking for is some understanding
regarding the complexity of the system we are seeking to understand and the
difficulty in the words we use. That instead of saying: no your are wrong,
no you do not understand, first make sure you have a shared understanding
of the terms being used. It may be a semantic problem, not a conceptual one.
Unfortunately, the problem is deeper than this. Let me add some more from
the posts of others.
Here is what i. Kurtzer said to Bruce A.
At the principle level you
still make comments that suggest you still haven't got it.
This was just the most recent instance.
Here is Bruce G. agreeing with i.
[From Bruce Gregory (980219.0526 EST)]
I whole-heartedly agree.
And, of course, our venerable Rick:
[From Rick Marken (980218.2050)]
While it is true that o = f(e) it is also (and simultaneously)
true that e = g(o). Saying that a control system "responds to
error" is a cause-effect description of the behavior of a
control system; it is equivalent to saying that a control system
"responds to perception", which is also incorrect.
and:
If one wants to say that a control system "responds to error"
one should also add (immediately) "while the error is responding
to the response to error". But I think it is much easier (and
clearer, becauase it calls attention to what is most important
about the control loop -- the fact that it keeps some perception
at a reference level) to say that a control system "controls
a perception".
and:
So that is why I corrected Jeff's description of the behavior
of a control system as "response to error". Besides being wrong
in a nit picky way, it is also wrong in an important way --
because it directs attention away from what is signifiant
about the behavior of a control system (controlled perceptions)
and toward what is superficial (the appearance of response
to error, which only occurs when there is a strong, transient
disturbance to a controlled perception).
Let me start with Rick. In the first part of the first sentence Rick
acknowledges that if Bruce A.s interpretation of what I said was correct
(which it was, pretty much), what I said was true. However, Rick cannot
let it go there. Rick has a point to make (I think the most accurate
representation of Rick's system is that the perceptions he creates about
whether others are understanding PCT and his points is very well specified,
allowing little to no variation in the feedback). His point is that PCT is
not just o = f(e), but that is must describe the whole loop. Usually, he
focuses on perception as the key term in an equation or sentence. The
second paragraph is saying the same thing it seems. Rick's problem is that
I was incomplete. What bothers me is that incomplete is equated with wrong
(see the last paragraph; see Bruce G. and i.). The perception I am
controlling for with these actions (posting this note) is that in the
sentence where Rick saids "it is also wrong in an important way" he instead
say, "it is also incomplete in an important way." Maybe I am nit picking.
But the tenor of the discourse is much different if that one little change
were made.
I see one of the central reasons for the current conflict is that this
point that Rick (and Bill and others) is trying to make is not the only
point of PCT. It is a central point. It is a major problem point for many
in psychology. It is largely, at least partially, responsible for a lot of
bad science. But it is not the only point.
Here I am, engaging in a behavior (action; output) to control a perception
which, if I am to believe Rick and Bill and others, can never happen. For
the perception is that someone (or ones) change because of my actions. The
perception is that the posts reflect an understanding of the nature of
control systems (that they are skeptical, that they have meanings for words
that might reflect a slightly different interpretation of what that word
meant by them in a previous post, etc.). I have never (okay rarely) seen
this perception in line with my reference signal. Yet, I continue to act.
In other words, e is not f(o), as my actions (outputs) have not affected my
errors. Yet, I continue to act. And, I continue to believe the PCT+ model
(due to Bill's recent post, the use of reorganization requires talking
about more the PCT). I believe, that because e not = f(o), that is,
because there is no correlation between e and o. I will reorganization
with my output or my input function until e = f(o), or the entire ECU goes
off-line (perhaps, gain = zero). In other words, to understand my
immediate behavior (am sorry, output), o = f(e) is a much more important
part of the set of equations than e = f(o). But that to understand my
eventually lack of output e = f(o) is important.
After this long post my message may be summarized in one word: chill.
After reading posts on this net for years, I was sure that when I suggested
to fellow psychologists that we control perceptions, they would think I was
crazy. Instead, I got "yes, that makes sense." It was almost like a
comedy routine where I would than say "No, you do not understand, the only
way we can control behavior is threw the perception of the results of that
behavior on the environment, which also has disturbances and requires
sensing and translating the environment into internal representations...."
But I stopped myself. I registered the feedback. They got it. What has
often happened here is that we are not always registering the feedback.
Some of us get it! You are making a difference. Please take the compliment!
Jeff
A great many people think they are thinking when they are merely
rearranging their prejudices.
-- William James
