--Boundary (ID 4YZBaLi+54nyMhXT1rigBg)
Content-type: TEXT/PLAIN
--Boundary (ID 4YZBaLi+54nyMhXT1rigBg)
Content-type: MESSAGE/RFC822
···
Date: Sun, 12 Feb 1995 11:30:00 MET
Subject: CSG-net--read this post second
Sender: James Wilk <WILK@WELL.SF.CA.US>
To: lcsmith <LCSMITH@restena.lu>
Content-type: TEXT/PLAIN
Posting-date: Sun, 12 Feb 1995 00:00:00 MET
Importance: normal
A1-type: MAIL
Lars,
As I told you on the phone, I tried to post the following to CSG-net last
Thursday, and I think it went to Tom Bourbon by mistake. I still can't
figure out how to reply to posts so that they end up on CSG-net itself.
Anyway, as per your request, here it is, and if you think it is still
relevant, and not to egregiously in error, you might want to forward it to
the gang:
To: tbourbon@HEART.MED.UTH.TMC.EDU
Subject: Re: Intro & Positive FeedBack ???
[James Wilk 950209.1737 GMT]
As this is my first attempt to post to CSG-net, and as my screen is ambiguous
at this point, I hope this gets there and not to you personally, Tom.
Here goes, anyway.
If two people look at a situation and one says, "There, _that's_ an
example of positive feedback!" and the other says, "But that's an example
of _negative_ feedback!" are they necessarily contradicting one another?
I would say to Tom and Lars (agreeing with Martin's comments especially),
there need be no contradiction at all. Let's take an example.
There is an old practical joke in which you arrange to introduce two of
your friends at a party, friends have never met one another. You say to
each of the two, "I'm about to introduce you to ________, who, I'm afraid,
has become rather deaf but refuses to wear a hearing aid, and is very
self-conscious about it; so whilst you mustn't let on that you know this,
you'll have to speak very loudly to be heard--louder than you imagine."
Having briefed them separately, you then politely introduce them and
retire to observe the unfolding spectacle. The two of course begin
speaking in slightly louder than average tones and each progressively
increases his own volume until the two are fairly bellowing at one
another, to the embarrassment and eventual delight of all around.
Now each of the two interlocutors can be thought of as embodying control
systems with negative feedback. Each is controlling for a volume level
loud enough to get through to the other, but no louder than _needs_ be.
They each try and achieve this by attempting, if you like, to _match_ the
volume level of the other. For (given the attribution of deafness) the
other fellow's volume level would normally provide a fairly reliable
guide, since the other fellow will (normally) be speaking in what he
fondly believes to be (what _sounds to him_ like) normal voice tone for
the environment--sc. at least loud enough for him to hear himself above
the din of the party.
The other person's volume thus sets the reference level for one's own
volume, and negative feedback ensures that one matches and sustains this
level. Under normal circumstances, when the surrounding noise level
momentarily increases, one has to speak louder to keep one's perceived
own-voice volume constant. And if the two victims of the joke are talking
"in full duplex", as they will be, and are trying occasionally to talk
over the other, the other person's voice will be another source of
perturbation as one attempts to control for the volume of one's own voice.
This of course is not normally consciously perceived, and even if it were
would be misattributed. Indeed the joke's success depends on the
continuing misattributions resulting from the initial disinformation.
This leads each speaker subliminally to attribute the other person's now
slightly raised voice _not_ to their need to talk over oneself (the
differences are usually too small to be consciously perceptible), but as
simply re-confirming the correct reference level for one's own voice
volume. Each probably _doesn't_ _even_ _recognise_ that this is in fact a
new level. (Try and "memorise" a volume level of someone else's voice
during a conversation--it's pretty nigh impossible with any degree of
precision.) And so every time a disturbance drives one voice level
slightly higher, a new, higher reference level for both of them is set.
In normal circumstances, at such a party, over a short time slice at any
rate, the tendency of the ambient perturbations (particularly those from
one's interlocutor) will be to drive the reference levels higher.
Besides, one usually wants to err on the side of caution and be heard; all
things being equal, one speaks perhaps a bit louder than one is absolutely
sure one needs to speak. (And the double disinformation ensures that the
false assumptions and resulting mis-attributions cannot be tested and
corrected.)
In some respects, this is merely an exaggerated version of what happens to
noise levels at parties anyway--why they tend to escalate over certain
intervals of time. And the well-known phenomenon of "an angel passing
over" (the room suddenly falling silent) occurs when a number of people
are trying to have private conversations in an environment that is getting
quieter, say because a few of the people in the restaurant are suddenly
busy eating, leading to a sudden drop in noise level, or because the Muzak
or air conditioning has suddenly cut out. Each individual is behaving as
a control system with negative feedback in modulating their voice tones,
but the overall effect, viewed at the level of the interaction as a whole,
is of positive feedback. The louder it gets the louder it tends to get,
and the quieter it gets the quieter it tends to get.
While we are on the subject of parties, there is the often remarked
phenomenon of oscillatory drifting between rooms. One room will fill up
with people drifting in from the other room, emptying the other room until
almost everyone is in one room, and then the drift will go back the other
way. Individuals may be controlling for getting to talk to the people
they want to talk to, and if those people are now to be found in the other
room, they'll have to go and find them there. And as they may also be
controlling for being in a room that isn't too crowded and smoky and
noisy--so a few will drift back into the quieter room and the drift begins
to go back in the other direction.
The "deaf friends" prank, the party noise volumes, the drifting between
rooms--are these instances of positive feedback or negative? As I
understand it, perhaps wrongly, they are equally instances of both.
Here's why I think so.
There is no paradox--WE SIMPLY ARE TALKING ABOUT DIFFERENT VARIABLES. To
see this, let's return to the auditory examples. First, there is the
variable "my own perceived voice volume." Each individual is controlling
for the volume of their own voice as they hear it against the background
of ambient noise, seeking to maintain it, say, at a constant level. Their
perception of their voice volume is able to have an effect on itself via a
closed loop, opposite to the direction of the feedback. For example, when
the other person is talking over oneself, or other voices increase in
volume, or increased stereo volume makes it harder to hear oneself, one
cancels out the effect of such disturbances by increasing one's own voice
volume. (Similarly, those who were listening to the music and are now
unable to hear it above the general din, turn up the music. And so on.)
But note that there is another variable we can equally well study, the
variable "overall volume of noise in the room". We can think of there
being a machine at work here (constituted by all the little negative
feedback systems taken collectively), a machine for increasing the noise
volume in the room. The louder the overall volume gets, the louder it
will _tend_ to get. For any disturbance increasing the overall volume
(say, experimentally turning on a vaccuum cleaner in the next room) will
raise the volume level by _more than_ the additional decibels contributed
by the Hoover. The higher the value of the variable, "overall ambient
noise level," the more this will have an effect on itself via a closed
loop, in the _same_ direction as the perturbations: positive feedback.
So if we suddenly turn off the music, or close the doors between
connecting rooms, the (summed) overall volume will drop, and the decreased
value of the variable "overall ambient noise level" will, in the way we
have seen (all those wonderful little control systems operating!), have a
closed loop (positive feedback) effect on itself until it reaches a limit
(here, total silence--"an angel passing over"). There is no contradiction
as long as we are clear about which machine we are studying and which
variables.
The symmetrical escalations ("schismogenesis" as Bateson called it)
characteristic e.g. of arms races are indeed capable of being viewed,
validly, as positive feedback phenomena, even though we can clearly see
how they come about through the interaction between control systems with
negative feedback. Each side tries to stay one step ahead of the other
(to have more warheads, for example).
Perhaps we ought to say (to remember?) that feedback is an _effect_ that
is observed, whether it is positive or negative; it is a contingent
relationship between variables. If disturbing a variable has an effect on
that variable in the same direction as the disturbance, then, by
definition, we are observing a positive feedback effect. If disturbing a
variable has an effect on that variable in a direction opposite to the
effect of the perturbation then we are observing a negative feedback
effect. One and the same disturbance may thus permit us to observe an
event which is at one and the same time an instance of a positive feedback
effect and of a negative feedback effect. Turning on the Hoover resulting
in everyone talking louder demonstrates the negative feedback effect on
each of the individual variables "my perceived voice volume" and
(consequently) a positive feedback effect on the variable "ambient noise
level".
As it happens, we can come to _discover_ that a positive feedback effect
is actually produced by underlying negative feedback mechanisms.
Discovering such things, and being able to demonstrate how one is a
consequence of the other, is part of what is known as "science".
Perhaps, Tom, this is near to what has been bugging you about Lars's
characterisation of the standard-establishing examples as being instances
of positive feedback. But in the tone of your remarks on the theme of
"show me the positive feedbackDDwhere is it, come on, where is it? _I_
don't see any positive feedback!" it comes across sounding rather
dogmatic, with almost religious zeal, as if the discovery of the
importance of the concept and phenomenon of negative feedback has now made
even the _recognition_ of the _phenomenon_ of positive feedback somehow
tantamount to apostasy. (Cries of "revisionism!") Like Martin (and Bill,
I think) I can see no conflict with PCT in characterising Lars's examples
as positive feedback. (After the debut performance of _The Huguenots_,
where Rossini and some of his pupils were in attendance, they eagerly
asked him, afterwards, "Maestro, maestro, what did you think of the
music?" "Music?" replied Rossini, "I didn't hear any music.")
Sorry for such a long post. Am I missing something here?
Best,
James
--Boundary (ID 4YZBaLi+54nyMhXT1rigBg)--