[From Rick Marken (980701.1030)]


It's clear that you don't want to call coercion "control of
behavior". Why, I don't know.

Bruce Nevin (980630.980701 1101 EDT) --

Because you are using "behavior" in the pre-PCT sense of external
observables available to the coercer (and the observer).

Not at all. I am using "behavior" in the post-PCT sense; behavior
is the control of perception. Post-PCT behavior is still externally
observable; you just have to know how to observe it. That's what
the Test is about. In the spreadsheet model the behavior of both
systems (qo, qo' and qi) was clearly observable. This is true in
the real world to. When you move a cat from the chair to the floor
you are watching a change in one of the cat's behavioral variables
(qo or qi).

You can't tell what a person is doing (their behavior) by
watching what they are doing

Right. But you _can_ tell what a person is doing by watching
"intelligently" -- through "PCT glasses". You watch for lack of
effect of natural or artificial disturbances on possible
controlled variables. When you do this you can usually see both
a person's actions _and_ the variable they are controlling with
those actions.

I guess you enjoy ridiculing people? What are you doing when you
do that? Is it getting you what you want?

I don't enjoy it. I have no idea what I'm doing (or when
I'm doing it) and it's not getting me what I want. Excuse me
while I slither off to the social skills room;-)

In actuality, I'm not clear about qi. It's the observer's
measurement of an aspect of the environment.

It's the observer's perception of a variable soneon else is
controlling. The distance between cursor and target in a
tracking task is qi.

I suspect that maybe qi is an artifact of observation and of the
protocols of experiment and modelling.

qi is a construct in the PCT model; it is also something an
observer can perceive if he knows what to look for.

I think maybe what is happening is that we're reifying qi as
though we had direct insight into Boss Reality.

Not at all. qi is a perception. In the model it's an environmental
variable but it only exists as a perception in an observer. Watch
the distance between cursor and target while a friend does the
tracking task at:

When you do this you are watching variations in qi.




Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail:

[From Rick Marken (950717.1000)]

The discussion of schedule data is very interesting and I hope that the PCT
account of these results can be published soon as a paper by Powers, Abbott
and Saunders. How about calling it "Operant behavior: The control of

I am somewhat surprised that there were no comments on my "Behavior and
Control" post (950714.1220). I still think that a good way to characterize
the difference between the PCT and conventional approaches to understanding
behavior is in terms of the definition of behavior itself; PCT views behavior
as controlled perceptual input; conventional psychology views behavior as
caused output.

In my earlier post, I said:

Most of the data of conventional psychology are measures of the visible side
effects of control; they tell us almost nothing about what organisms are
actually doing (controlling). This is what my "mind reading" demo is

I was hoping that Bruce Abbott (who, though unsympathetic to conventional
models of behavior -- such as reinforcement theory -- seems sympathetic
to the conventional approach to understanding behavior) would comment on that
demo. As I said, the demo seems (to me) to show quite clearly why it is
inappropriate to deal with behavior as it is delt with in conventional
psychology. The mind reading demo creates a situation you can't tell what a
person is doing (controlling) by looking at their behavior (results of

The only way to know what the person is actually doing in any situation is
by determining what s/he is controlling. Conventional approaches to
understanding behavior are never based on a systematic attempt to determine
what the subject is controlling (though data are sometimes available -- as it
is in the operant scheduling studies -- that makes it possible to do ad hoc
determination of a controlled variable).

What do you think of the "mind reading" demo, Bruce? I have presented it to
several conventional psychologists. They all found it "fun" but none dropped
what they were doing to start "following PCT". What is your take on this
demo? I am really curious about why it hasn't made more of an impression.



[From Fred Nickols (970913.0740 EST)]

Fred Nickols (970911.2240 EST)

I'm a one-time behaviorist and I never--not once--viewed
behavior as output... To me, behavior was and is a shorthand
label for "the activity of the organism."

Rick Marken (970911.2250)

Same thing. The visible activity of an organism is what we
call its "output". In PCT, the activity of the organism is
a side-effect of the process of perceptual control.

Okay; what about covert behavior? How does that fit with the
notion of behavior as output?


Fred Nickols

[From Bill Powers (970913.l0854 MDT)]

Fred Nickols (970913.0740 EST)--

To me, behavior was and is a shorthand label for "the activity of the
organism." [rather than outputs].

I presume that as a behaviorist you would mean "the observable activity of
the organism." All the observable activity of the organism is produced by
motor outputs and glandular outputs. Are there any exceptions to that?

What, to you, distinguishes behaviorism from any other branch of peychology?

Okay; what about covert behavior? How does that fit with the
notion of behavior as output?

One of the nice things about making up your own theory is that you don't
have to use other people's language. In PCT there is no such thing as
covert behavior (excepting a few things like breathing and swallowing,
which occur outside the nervous system but inside the body). In PCT,
behavior is strictly the effect that nervous system outputs have on
physical variables, usually via the muscles. Most of these effects are
visible from outside the organism. If we're just talking generally about
what an observer sees going on, we call it "behavior". If we mean something
more specific, we will say "action," meaning the immediate cause (like a
muscle contraction) of an observable behavior. If we mean a particular
measure of output that we would use in an analysis or a model, we would say
"output quantity" meaning a measurable physical quantity like a force.

Inside the nervous system there are no behaviors; there are only varying
signals. Of these signals, according to the theory as it is at present, the
only ones that we consciously experience are perceptual signals. If
anything is to be experienced of other activities inside the brain, it must
somehow be converted to a signal in the perceptual input channels. The
imagination connection was provided so that reference signals and past
values of perceptual signals (stored in memory) can be experienced. If you
wanted error signals to be experienced, you would have to propose another
special connection so that a copy of an error signal could get into a
perceptual input channel. You would also have to show that the system could
operate with such a special connection in place, just as a working model. I
haven't proposed the latter connection because I can't make a model
containing that connection work in simulation.


Bill P.

[From Rick Marken (970913.0940)]

Fred Nickols (970913.0740 EST) --

what about covert behavior? How does that fit with the
notion of behavior as output?

From the "behavior as output" perspective, covert behavior is

just output that you (the observer) can't see. From the
"behavior as controlled input" perspective, covert behavior is
controlled imagination.

Point at this "X" on the screen; that's overt behavior -- control
of the perception of of the relationship between your finger and
the "X". Now close your eyes and imagine pointing at the "X" on
the screen. That's covert behavior -- control of the imagined
perception of your finger relative to the screen.

Note, by the way, that an observer of your overt behavior would
not be able to tell what you are _doing_ (controlling the
perceived relationship between finger tip and "X") just by looking
at your "behavior". The observer might notice that you are lifting
an arm, pointing a finger at the computer screen, pointing at any of
the letters in the vicinity of the "X", making a 90 degree
angle between your arm and your body, etc. All these would be
perfectly accurate descriptions of what the observer sees you
doing. But it is highly unlikely that the observer would notice
what you are _actually_ doing (from your point of view); that you
are controlling the perceived relationship between your finger
tip and the "X".

The only way for an observer to find out what another organism
is actually doing (what perceptions it is controlling) is by
doing The Test for the Controlled Variable. You have to guess
at the variable being controlled, apply disturbances to that
variable and look for _lack_ of effect; if the disturbances are
effective then you have to revise your guess about the variable
under control, apply disturbances to this variable and, again,
look for lack of effect. You iterate this process until you
discover a variable (like the relationship between finger tip and
"X") that is _not_ affected by disturbances; for example, you move
the "X" around the screen (disturbance) and find that the
relationship between finger tip and "X" is not affected; the person
changes pointing angle to keep the finger tip pointing at the "X".

If you don't Test to determine what people are controlling then
you have no idea what they are doing; when you study behavior
without Testing for Controlled Variables all you are doing is
studying what _you_ think a person is doing; you are studying
the behavioral appearances that _you_ find interesting; these
appearances (like the 90 degree angle you make when pointing
at the "X") are likely to be only incidentally related to what
the actor is actually doing (controlling).

Conventional psychological researchers have no idea that the
behavior they see and find interesting may have little or
nothing to do with what the actor is actually doing.
Psychologists study behavior that looks interesting to them.
The only objective criterion that I have ever heard of for
selecting one aspect of behavior for study over another is
quality of results; you study the behavior that gives you
the best data (which is usually pretty vad anyway). There is
no guarantee that behavior selected by this criterion will be
anything other than a side effect of what a person is actually
doing (controlling).

Since psychologists have never (intentionally) looked at behavior
in terms of the variables organisms control there is very little
existing behavioral research that can be used to test PCT.
This is one reason why the perennial debate about reinforcement
is so fruitless; the data relevant to a PCT analysis of
operant behavior (including learning) have never been collected.

People who don't understand the nature of behavior (as controlled
input) don't think that Testing for Controlled Variables is
particularly important; they see it as just an esoteric aspect
of PCT; a peculiar, unfamiliar and unnecessary approach to doing
behavioral research. They assume that psychological researchers
have been doing things right all along and that PCT should
account for existing psychological data. I hope you can see now
that this is simply not the case; existing psychological data
are as likely to be relevant to what organisms are actually
doing (controlling) as any particular description of your
"pointing behavior" is likely to be relevant to what you are
actually doing (pointing at the "X").

Understanding the PCT view of behavior is, I think, more
important than undertstanding the PCT model. The PCT
revolution is a revolution in the way we see behavior;
behavior is not the pattern of activities that we see;
behavior is controlled perception; what we see is either
a side effect of the process or an observable correlate of what
is actually being controlled; The Test is how we discriminate
the former from the latter.

You can see all these ideas illustrated in living black and white
in my "Mind Reading" demo (called "The Test for the Controlled
Variable) at:





Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: