The behavioral illusion

CSGnet Archive CSG1997
1

[From Bruce Abbott (971208.2015 EST)]

The "behavioral illusion" occurs when a control system is successfully
controlling some variable, keeping it near the system's reference level by
producing output that opposes the effects of disturbances to the controlled
variable (CV). If control were perfect (assuming a constant reference and
that the CV has already been brought to equilibrium with the reference), the
effects of disturbance and output on the CV would be equal and opposite, so
that the two influences cancel exactly, leaving the CV unchanged at some
value close to that of the reference. Result: the effect of output on CV is
a mirror image of the effect of disturbance on CV.

Note that it is not the output of the control system that mirrors the effect
of the disturbance, but the effect of this output on the CV. If the
"environmental feedback function" that relates system output to effect on CV
changes, then output will have to change to bring the effect of output on CV
back to its previous level, so that it once again cancels the effect of the
disturbance. For example, if the effect of output on CV is cut to 1/3rd its
previous value, output must rise by a factor of 3 in order to restore the
same opposition to the disturbance.

As you examine the effect of disturbance on output (with a given feedback
function), you are not learning much of anything about the control system
itself, but you _are_ learning how you can control the output of the system,
so long as the system continues to control the CV in the same manner and
hold the same reference value. For example, so long as the person on the
other end of that rubber band continues to attempt to keep the knot over the
coin by the rules established, you can control where that person's end of
the rubber band will be. If one can also establish conditions such that the
person will want to continue controlling the knot's position at the coin
(perhaps by offering a large amount of money for doing so, if that will work
for this person), then one will be able to control the person's rubber band
pulling behavior rather accurately. For someone interested in learning how
to establish control over another's behavior, it might make little practical
difference that the observed variations in behavior reveal only the inverse
of the environmental feedback function.

Another case in which the behavioral illusion may be of value occurs when
the "environmental" feedback function consists of elements located within
the organism. If the effect of the output on the CV cannot be directly
observed from the outside, the form of the function could in principle be
deduced _if_ the reference level is held constant over the period of
observation _and_ the CV is being well controlled. Or so I surmise. Does
that sound reasonable? If so, then something of value about the organism
could be learned from the behavioral illusion after all, no?

Regards,

Bruce

2

[From Bill Powers (971209.0232 MST)]

Bruce Abbott (971208.2015 EST)--

The "behavioral illusion" occurs when a control system is successfully
controlling some variable, keeping it near the system's reference level by
producing output that opposes the effects of disturbances to the controlled
variable (CV).

A good discussion of the basic behavioral illusion.

As you examine the effect of disturbance on output (with a given feedback
function), you are not learning much of anything about the control system
itself,

Except, of course, that it _is_ a control system rather than a
straight-through system...

but you _are_ learning how you can control the output of the system,
so long as the system continues to control the CV in the same manner and
hold the same reference value.

Another condition is that the control action required not _cause_
significant errors in other control systems of the same or higher level.
While this is covered under "continues to control the CV...", this point is
often overlooked. Your ability to control the other systems's output this
way depends on (a) making sure the disturbance is not large enough or fast
enough to prevent control of the CV, and (b) making sure the action
required has no important side-effects on anything else the other system is
controlling. Thus you will gladly work to get money, but not if the work is
painful, exhausting, or disruptive of your family life (unless you have no
choice).

For example, so long as the person on the
other end of that rubber band continues to attempt to keep the knot over the
coin by the rules established, you can control where that person's end of
the rubber band will be.

Right, but don't expect it to work if you run the other person's hand into
a hot soldering iron.

If one can also establish conditions such that the
person will want to continue controlling the knot's position at the coin
(perhaps by offering a large amount of money for doing so, if that will work
for this person), then one will be able to control the person's rubber band
pulling behavior rather accurately.

This demonstrates a two level control process: a person using a system that
controls one variable as a means of controlling another. If I can get a
large amount of money by keeping a knot over a dot, I may well do so. In
fact, if this is a legitimate offer and the amount of money is large enough
to cover air fare and expenses to and from Indiana, plus a little, I will
be happy to take you up on the offer, say the first week in January? How
about it? I'll be happy to let you control my rubber-banding behavior for,
say, three hours, by giving me, say, $1500. In fact I'll give a seminar for
your students and consult with you for one day on your rat data if you'll
round it up to $2000. I'm sort of saving up for a laptop with a bigger
screen and a CD ROM, so we'd both get something we want -- you get to
control some of my behavior, and I get closer to my new laptop. Sound like
a good deal? How far into this gig of controlling other people's behavior
are you? How does $1999.95 sound? This offer may be withdrawn at any time,
prices are subject to variation. Remember, Christmas is coming. I will give
gift certificates if you want to let someone else control me.

For someone interested in learning how
to establish control over another's behavior, it might make little practical
difference that the observed variations in behavior reveal only the inverse
of the environmental feedback function.

Right, that's what I'm saying, too. For only $1999.95, you get to control
the rubber-banding behavior of William T. Powers for three whole hours.
I'll even sign any affadavits you want to show to your colleagues to prove
you actually did it by reinforcing my behavior. Do you think any of them
might be interested in a similar opportunity? (No BDSM, though -- rubber
bands only, in the standard position. I am not kinky. I don't do levers).
Is there a significant market for this sort of thing?

Another case in which the behavioral illusion may be of value occurs when
the "environmental" feedback function consists of elements located within
the organism. If the effect of the output on the CV cannot be directly
observed from the outside, the form of the function could in principle be
deduced _if_ the reference level is held constant over the period of
observation _and_ the CV is being well controlled. Or so I surmise. Does
that sound reasonable? If so, then something of value about the organism
could be learned from the behavioral illusion after all, no?

Right, that's part of what I call the Test. One day, given some resources
(say, $199,950), we might be able to work out a multiple-level Test, in
which we keep track not only of a motor action and one controlled variable,
but other variables that are controlled by varying the reference level for
the first controlled variable. By solving multiple equations we could
establish the changing reference levels as well as the changing actions as
part of a multi-system control process. If you know of anyone who is really
hot to control the behavior of a PCT researcher, I'm sure we could work
something out -- if not with William T. Powers, then with someone else I
could get to carry out the project, for, say, one percent of the total take
-- er, reward. That would come to only $1999.95.

Honi soit qui mal y pense,

Bill P.

P.S. The behavioral illusion is an illusion only if you interpret the
disturbance as a stimulus that acts via the organism to produce a response,
and assume that it reveals the organism function. It's the same as with any
illusion; the plumber's illusion or the Ames Window illusion is an illusion
only if you are fooled into taking appearances as a true picture of nature.

3

[From Bruce Abbott (971209.1615 EST)]

Bill Powers (971209.0232 MST) --

Bruce Abbott (971208.2015 EST)

If one can also establish conditions such that the
person will want to continue controlling the knot's position at the coin
(perhaps by offering a large amount of money for doing so, if that will work
for this person), then one will be able to control the person's rubber band
pulling behavior rather accurately.

This demonstrates a two level control process: a person using a system that
controls one variable as a means of controlling another. If I can get a
large amount of money by keeping a knot over a dot, I may well do so. In
fact, if this is a legitimate offer and the amount of money is large enough
to cover air fare and expenses to and from Indiana, plus a little, I will
be happy to take you up on the offer, say the first week in January? How
about it? I'll be happy to let you control my rubber-banding behavior for,
say, three hours, by giving me, say, $1500. In fact I'll give a seminar for
your students and consult with you for one day on your rat data if you'll
round it up to $2000. I'm sort of saving up for a laptop with a bigger
screen and a CD ROM, so we'd both get something we want -- you get to
control some of my behavior, and I get closer to my new laptop. Sound like
a good deal?

The point is, Bill, that influencing another person to behave as one wants
him or her to behave presents no insuperable difficulties, and in fact, PCT
tells me exactly what conditions have to hold if my offer is to produce the
outcome I wish it to. But you are no helpless pawn in this; you can
negotiate for a better deal, as you do in your example, thereby influencing
_my_ behavior.

How far into this gig of controlling other people's behavior
are you? How does $1999.95 sound? This offer may be withdrawn at any time,
prices are subject to variation. Remember, Christmas is coming. I will give
gift certificates if you want to let someone else control me.

Actually, I personally am not into controlling other people's behavior. But
I'd love to have you come for a visit; I'll even pick you up at the airport.
(:->

For someone interested in learning how
to establish control over another's behavior, it might make little practical
difference that the observed variations in behavior reveal only the inverse
of the environmental feedback function.

Right, that's what I'm saying, too. For only $1999.95, you get to control
the rubber-banding behavior of William T. Powers for three whole hours.
I'll even sign any affadavits you want to show to your colleagues to prove
you actually did it by reinforcing my behavior. Do you think any of them
might be interested in a similar opportunity? (No BDSM, though -- rubber
bands only, in the standard position. I am not kinky. I don't do levers).
Is there a significant market for this sort of thing?

Well, if you don't do levers, there's no point, is there? Anyway, I could
probably get Jane Fonda for around that price (she used to command more, but
those days are gone) and she's better looking than you are, if you'll pardon
my saying so.

Regards,

Bruce