Questions about Behavior

[Fred_Nickols3]

[Fred Nickols (970914.1110 EST)]

    >>Fred Nickols (970913.0740 EST)

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

  >Bill Powers (970913.l0854 MDT) [ <-- What time is that, Bill? ( ]

  >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?

Being completely lacking in formal, college-level academic credentials, I
don't know that I can rightly lay claim to the title of "behaviorist" except
in my own eyes, so I'll just stick with what I mean by behavior, sans any
titles or philosophical affiliations.

Your question about exceptions could apply to your modification of my
definition of behavior or to your statement about motor outputs and
glandular outputs. I'll answer each...

If by "observable" you mean both visible to the naked eye AND that which is
"observable" via other means (e.g., EEG, EKG, etc.), then I would agree that
both are encompassed by "the activity of the organism." However, that is
not all that I mean by behavior as "the activity of the organism." Also
included in my definition of behavior is that activity which John Watson
onced called "sub-vocal speech" (i.e., verbalizations not spoken aloud) and
which, in my view, comprise a great deal of what is ordinarily meant by
"thinking." I'm not up enough on biology to know if that kind of quiet
verbal activity is thought or proven to be produced by glandular or motor
outputs, but there are times when I'd swear it's motor activity that's been
cut off just short of utterance.

As for the second application of your question about exceptions, I don't
know if motor and glandular activity account for all activity or not. Where
do neurons firing fit in? Are they the products of glandular activity? If
so, I'd guess that motor and glandular activity do account for all of the
activity of the organism.

  >Bill Powers (970913.l0854 MDT)

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

That's an easy one: The commitment to explaining human behavior based
solely on what can be observed. Most of the dyed-in-the-wool (and suitably
credentialed) behaviorists I've known were steadfast in their view that we
do not and cannot know what is going on inside the organism with respect to
concepts like purpose, intent, goals, expectations, etc. These are all
self-reported and therefore not reliable. So, they are not part of the
behaviorists' scheme of things. I don't know any behaviorists who would
deny that such phenomena exist; instead, the behaviorists I know simply
eschew any speculation about such internal mechanisms and focus instead on
the external, observable phenomena.

Questions answered?

  >Bill Powers (970913.l0854 MDT)

  >I can see from the rest of this post that you're still struggling with the
  >reasoning here. All I can say is, keep going back to the basic fact:
  >behavior varies from one situation to another, even if the consequence it
  >produces remains the same. Therefore behavior is not a function of the
  >consequence. The explanation of the observations does not lie in any aspect
  >of the environment. The right way to state the observations is that the
  >organism will vary its behavior in any way needed to produce the
  >consequence it wants.

Heck, I've been struggling with reasoning for as long as I can remember; not
much new there--and I'm not sure I'm getting any better at it. However, I
think I have a grip on the essence of one piece of the apparent conflict.

I agree with the notion that behavior varies from situation to situation to
maintain some kind of constancy in outcomes or consequences. I also agree
with the notion that the organism will vary its behavior in any way needed
to produce the consequence it wants. Where the rub seems to be is that I
also accept the statement that behavior is a function of its consequences.
To you, that statement carries formulaic content [b=f(c)] and in that
formula "f" refers to "function" in the sense of some fixed relationship.
I, too, have used b=f(c) but only in the general sense that behavior depends
in some unspecified way on its consequences. Obviously, my behavior in the
here and now can't depend on consequences that are off in the yet to occur,
but I am convinced that my behavior depends very much on the kinds of
consequences I envision it will produce. That, I believe, is consistent
with your own view. The envisioned consequences of my behavior are very much
a product of past experience.

Would it be more precise and acceptable to say, "Current behavior is
influenced by remembered perceptions of relationships between behavior and
its consequences."
Regards,

Fred Nickols
nickols@worldnet.att.net

[R_Marken]

[From Rick Marken (970914.1920)]

Fred Nickols (970914.1110 EST) --

I also accept the statement that behavior is a function of its
consequences...Obviously, my behavior in the here and now can't
depend on consequences that are off in the yet to occur,
but I am convinced that my behavior depends very much on the
kinds of consequences I envision it will produce. That, I
believe, is consistent with your own view.

The idea that one's behavior depends on the kinds of consequences
_envisioned_ is indeed, perfectly consistent with PCT. _Envisioned
consequences_ are called wants and desires; in PCT they are
called reference signals or reference perceptions. They are
the consequences that one _intends_ to produce; behavior (actions)
depends very much on what consequences are envisioned (intended);
it also depends on prevailing disturbances. We say it this way:

qo = r - d

Variations in behavior (qo) depend on variations in reference (r)
specifications for controlled perceptions and variations in
disturbances (d) to those perceptions.

Saying that behavior depends on _envisioned_ consequences is a
_lot_ different (and a lot more correct) than saying that behavior
depends on its consequences. My behavior does depend on the fact
that I envision certain consequences, like eating the nice
sirloin waiting for me in the fridge. But my behavior does not
depend on the fact that a sizzling sirloin is a consequence of
my actions (as I hope it will be). This is the mistake made by
those who believe in reinforcement; they believe that behavior
actually depends on the occurrance of certain (reinforcing)
consequences (like food), These consequences are actually
envisioned (wanted) and produced by behavior (actions); they
are controlled variables.

Best

Rick

···

--

Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken/

[Bruce_Abbott]

[From Bruce Abbott (970915.1035 EST)]

Rick Marken (970914.1920) --

Fred Nickols (970914.1110 EST)

I also accept the statement that behavior is a function of its
consequences...Obviously, my behavior in the here and now can't
depend on consequences that are off in the yet to occur,
but I am convinced that my behavior depends very much on the
kinds of consequences I envision it will produce. That, I
believe, is consistent with your own view.

The idea that one's behavior depends on the kinds of consequences
_envisioned_ is indeed, perfectly consistent with PCT. _Envisioned
consequences_ are called wants and desires; in PCT they are
called reference signals or reference perceptions. . . .

Rick, this is something I'm not clear on. Is a reference signal the same as
a reference perception? If not, what's the difference? Under PCT, are
reference signals perceived?

Saying that behavior depends on _envisioned_ consequences is a
_lot_ different (and a lot more correct) than saying that behavior
depends on its consequences. My behavior does depend on the fact
that I envision certain consequences, like eating the nice
sirloin waiting for me in the fridge. But my behavior does not
depend on the fact that a sizzling sirloin is a consequence of
my actions (as I hope it will be). This is the mistake made by
those who believe in reinforcement; they believe that behavior
actually depends on the occurrance of certain (reinforcing)
consequences (like food).

Where does the rat get its _envisioned_ consequence? That is, how does it
know what the consequence of its actions will be?

Regards,

Bruce

[Bruce_Gregory9]

[From Bruce Gregory (970915.1210 EDT)]

Bruce Abbott (970915.1035 EST)]

>The idea that one's behavior depends on the kinds of consequences
>_envisioned_ is indeed, perfectly consistent with PCT. _Envisioned
> consequences_ are called wants and desires; in PCT they are
>called reference signals or reference perceptions. . . .

Rick, this is something I'm not clear on. Is a reference signal the same as
a reference perception? If not, what's the difference? Under PCT, are
reference signals perceived?

Let me try while we await Rick's response. Reference signals
are not perceived. Remembered perceptions, however, can serve as
reference signals. This might be what Rick is referring to.

>Saying that behavior depends on _envisioned_ consequences is a
>_lot_ different (and a lot more correct) than saying that behavior
>depends on its consequences. My behavior does depend on the fact
>that I envision certain consequences, like eating the nice
>sirloin waiting for me in the fridge. But my behavior does not
>depend on the fact that a sizzling sirloin is a consequence of
>my actions (as I hope it will be). This is the mistake made by
>those who believe in reinforcement; they believe that behavior
>actually depends on the occurrance of certain (reinforcing)
>consequences (like food).

Where does the rat get its _envisioned_ consequence?

From earlier experiences.

That is, how does it
know what the consequence of its actions will be?

It doesn't, but it knows what it wants to perceive.

bruce

[R_Marken]

[From Rick Marken (970915.0930)]

Bruce Abbott (970915.1035 EST)

Rick, this is something I'm not clear on. Is a reference signal
the same as a reference perception? If not, what's the
difference?

Excellent questions!!

We have used "reference perception" and "reference signal"
synomymously; both refer to the specification, in a control loop,
for what the perceptual input _should be_. But I think the
term "reference perception" is ambiguous; it can be thought of
as referring to the reference signal (the _intended_ state of
a perception) or to the perception _when it is in the
reference state_. I think it would be clearer if we always
used the phrase "reference for the perception" as the synonym
for "reference signal" and "reference state of the controlled
perception" to refer to the controlled perception when it is in
it's reference state.

I could then describe my example of the sirloin this way: my
"reference for the perception" (my reference signal) was my
desire to perceive "steak". At the time I wrote the example I
wasn't perceiving "steak"; my perception was not in its reference
state. But, shortly after posting, I went in the kitchen and took
action (appropriate to prevailing circumstances) that produced
the perception of a sizzling sirloin: that perception was the
"reference state of the controlled perception".

Under PCT, are reference signals perceived?

In PCT, the only way a reference signal (reference state for a
perception) is perceived is when it is played back into the
control loop as a perception; this is the imagination connection.
So, when I was thinking about the sirloin I wanted to perceive
(the reference signal) -- and was not yet perceiving -- I was
talking about an imagined perception -- the "sizzling sirloin;
the "reference state of the controlled perception". But it's not
really the "referencesignal" that we perceive (imagine); it is
the reference state of the controlled perception; the state of
our perception that would be exactly right (match the reference
signal) if it occurred.

Ordinarily, I think our reference signals (wants) are _not_
perceived via imagination; they just sit silently in the
background, determining whether what we are perceiving is
right or wrong. I think we start becoming aware of our
reference signals (via imagination) when our experiences start
to differ substantially from what we want (for whatever reason;
loss of control or sudden, overwhelming disturbace). I also
think it is quite possible for people to be (and remain) quite
unaware of their reference signals (what they want). One goal of
therapy, I think is to make people aware of what they want and,
more important, _why_ they want it.

Where does the rat get its _envisioned_ consequence?

This is the "why do they want it" question; why does the rat
envision recieving food pellets at a particular rate? I think
you've already answered that; from the HPCT perspective it's
because envisioning (wanting) that result allows some other
control system (such as the nutrient control system) to achieve
the results that it envisions (wants).

That is, how does it know what the consequence of its actions
will be?

A control system doesn't "know" what the consequence of its
actions _will_ be; it gets a specification (in the form of a
reference signal) for what the consequences of its actions
_should_ be. If it is a good control system, and there are no
insuperable disturbances (like an experimenter who cuts off the
food supply) the consequences of its actions _will_ be what are
specified by the reference signal (what they _should be).

Best

Rick

···

--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken

[Bruce_Gregory9]

[From Bruce Gregory 9970915.1400 EDT)]

Rick Marken (970915.0930)

I _think_ my responses to Bruce are consistent with yours.
Please tell me if I am off base.

I also
think it is quite possible for people to be (and remain) quite
unaware of their reference signals (what they want). One goal of
therapy, I think is to make people aware of what they want and,
more important, _why_ they want it.

What do you have in mind as an appropriate answer to the
question "Why?" in a therapeutic situation? Is it a story about
how you established a certain reference level for a perception?
How is such a story empowering?

bruce

[R_Marken]

[From Rick Marken (970915.1250)]

Bruce Gregory (9970915.1400 EDT) --

I _think_ my responses to Bruce are consistent with yours.
Please tell me if I am off base.

Perfectly consistent. You use the term "remembered perceptions"
to describe what I would call "imaginations". Either term
works for me; what we are talking about are reference signals
that are played back into a perceptual function, comme ca:

                      r2
                       >
                  p2--|c|---e
                   ^ |
                   > >
                  >I> >O>
                   ^ |
                   > r1
                    \______/

I've drawn a level 2 control system whose output becomes the
reference (r1) for a level 1 system. In the diagram the level
1 reference has been short circuited into the input function
of the level 2 system. The level 2 system is in imagination
mode; it is getting exactly the perception (p2) being requested
(by r2); if consciousness is directed at p2, you become
aware of what you want; p2 is a "remembered" or "imagined"
perceptual correlate of the want (r2)

What do you have in mind as an appropriate answer to the
question "Why?" in a therapeutic situation?

Sort of. I wouldn't expect an explit answer like "I want this
because...". I think it is always possible to make up plausible
answers to the "why" question that really have nothing to do
with what's really going on. What I have in mind for therapy is
a change in conscious point of view. I think therapy occurs
when a person can look at problem goals _from the point of
view_ of the systems that are selecting these goals; from
the point of view of the "why" systems.

I understand this "point of view" thing in terms of the "rubber
band" conflict that is discussed in Bill's new book, "Starting
Over". As long as you look at the rubber band situation from
the point of view of trying to keep the knot on target and, at
the same time, trying not to break the rubber bands, you remain in
conflict; always experiencing some error (pain). But when you look
at the situation itself -- as an impossible situation -- then you
can select another situation that doesn't create all the pain; you
just change your goal about being in the rubber band situation.
What you have done is "go up a level" to see the conflict from
the perspective of the systems that selected playing along with
the rubber band game in the first place; these systems can select
not playing along anymore, thus eliminating the conflict. But you
can't change your "situation" goals while you are busy trying to
achieve the "knot on target"and "intact rubber bands" goals that
define the selected situation.

Best

Rick

···

--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken

[Myrexsw]

i.kurtzer (970915.2200)

[From Bruce Gregory (970915.1210 EDT)]

Reference signals
are not perceived. Remembered perceptions, however, can >serve as
reference signals.

Let us be clear about the subject matter we are referring to--system
analysis or phenomology- as
A) the answers are different
B) we should not talk mush
C) we cannot restrain PCT to introspection correspondances, though their
correspondance is the closest i've seen between these two types of
considerations.

Lastly, why don't those interested do something other than talking about
party lines and get off your asses and experiment.

still pissed at half-assed sincerity

i.

[HANS_BLOM2]

[Hans Blom, 970916]

(Rick Marken (970915.1250))

what we are talking about are reference signals that are played back
into a perceptual function, comme ca:

                      r2
                       >
                  p2--|c|---e
                   ^ |
                   > >
                  >I> >O> the controller
                   ^ | -------------------
                   > r1 its imaginary world
                    \______/

This model of imagination seems to pose that the imaginary world of a
controller delivers an output (to |I|) identical to its input (from

O>). Is this just a particular example or do you more generally

propose that it would be fitting (even if only in imagination
that we, in our imagination, represent the world by the equation
output = input or the controller, in its imagination mode, by the
equation perception = action? If so, what is the relation between
that imaginary world and the real one?

Greetings,

Hans

[R_Marken]

[From Rick Marken (970916.0830)]

Hans Blom (970916) --

do you...propose that it would be fitting (even if only in
imagination that we, in our imagination, represent the
world by the equation output = input

No. In imagination, the perceived world is

p2' = f(r1)

Where p2' is an imagined level 2 perception, r1 is the reference
input to the lower level system that would ordinarily be responsible
for producing the lower level perceptual input (p1) to f() and f()
is the level 2 perceptual function. Under ordinary circumstances
(non-imagination mode)

p2 = f(pl)

When level 1 control is nearly perfect, p1 = r1, so what you
perceive is what you would imagine in imagination mode; p2' = p2.

what is the relation between that imaginary world and the
real one?

By "real one" I presume you mean "perceived one". I just described
it above. The imaginary world is the same as the perceived world
to the extent that all lower level control is possible and perfect.
You can imagine waving your arms and flying but you can't perceive
it because you cannot produce the lower level perceptions (of
lift, mainly) that would produce this perception in fact.

Hans Blom (970916b) to i.kurtzer (970915) --

I'm very much in favor of it. But maybe I'm more in favor of
the search than of the truth

Then you have a most bizarre way of searching since you completely
ignore experimental evidence. You seem like a kid playing the
"hot" and "cold" while _intentionally_ ignoring all the cries of
"hot" and "cold" (the experimental evidence) around you because
you _know_ that the truth is hidden in that great big "model-based
control" box.

i.kurtzer (970915) --

Are you suggesting that a monkey knows about tensile strength as it
swings on a vine?

Hans Blom (970916d) --

Certainly.

I thought it was the search for truth that mattered to you. But
here you say you are _certain_ that this highly unlikely
proposition is true. Have you done experiments to show that
this is true. Have you even tried the experiment

http://home.earthlink.net/~rmarken/ControlDemo/OpenLoop.html

that shows that people (maybe not monkeys) do not use a model of
the physical connection between themsevles and the environment to
control?

Best

Rick

···

--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken

[Bill_Powers1]

[From Bill Powers (970917.1545 MDT)]

Hans Blom, 970916--

[Rick's diagram]

                     r2
                      >
                 p2--|c|---e
                  ^ |
                  > >
                 >I> >O> the controller
                  ^ | -------------------
                  > r1 its imaginary world
                   \______/

Hans' comment:

This model of imagination seems to pose that the imaginary world of a
controller delivers an output (to |I|) identical to its input (from

O>). Is this just a particular example or do you more generally

propose that it would be fitting (even if only in imagination
that we, in our imagination, represent the world by the equation
output = input or the controller, in its imagination mode, by the
equation perception = action? If so, what is the relation between
that imaginary world and the real one?

As Rick pointed out in his reply, this imagination connection has some
subtle properties that are not visible at a glance. Normally, when a
higher-level system sends its output to serve as a reference signal for
lower-level systems, the result is a set of upgoing perceptual signals from
the lower systems that contribute to the higher-level perception. When the
outgoing reference signal is substituted for the lower-level perception,
the effect is just as if the lower systems had instantly and perfectly
acted on their environments to produce the requested perception.

The motivation for adding this connection to the model was to account for
exactly the same phenomenon in experience. When you imagine how to fasten
two pieces of wood together, you might think of using an industrial staple
gun. In imagination, you fasten the pieces together in a jiffy, just as if
you owned such a staple gun and knew how to use it. You say, "Yeah, that
will work and I'll be done before supper time." You can do this even if you
have only a vague idea of what a staple gun is, and don't actually have
one. Basically, you imagine the experience of having and using one, but
only at the level of "fastening the pieces together." None of the
lower-level details is actually present.

When you then use the same reference signal as a reference input to the
real lower-level systems that have to be involved, you discover that they
do NOT immediately return the requested perception of a staple gun
fastening the pieces together. For one thing, you have to find out where to
buy a staple gun, purchase it, and learn how to use it. Then, when you do
try to use it, you find that the staples are too short to go through both
pieces of wood, and anyway that they're much too weak for the job. What
worked in imagination doesn't work nearly as well when you set the same
reference condition in the real-time mode. You certainly won't get the job
done by supper time.

In the imagination mode, the properties of the world at the lower levels
don't have to be considered. You imagine a staple gun, and there it is in
imagination, immediately. You can then consider the result and see if it's
what you want, just as if it had really happened. If there's something
about the imagined result you don't like (maybe you imagine shooting a
staple into your hand), then the lower-order problems become moot -- you
never have to buy the staple gun, learn to use it, and so on.

That's the sort of thing the imagination connection was supposed to
accomplish.

Best,

Bill P.

[HANS_BLOM2]

[Hans Blom, 970918c]

(Bill Powers (970917.1545 MDT))

[Rick's diagram]
                     r2
                      >
                 p2--|c|---e
                  ^ |
                  > >
                 >I> >O> the controller
                  ^ | -------------------
                  > r1 its imaginary world
                   \______/

As Rick pointed out in his reply, this imagination connection has
some subtle properties that are not visible at a glance.

You can say that again! Take a closer look at this part:

                 >I> >O> the controller
                  ^ | -------------------
                  p a its imaginary world
                   \______/

where I have renamed r1 into a (the "action" of the higher-level
controller) and added the symbol p (the "perception" offered to the
input of the higher-level controller. So for this higher-level
controller we have p=a: it perceives (only) its action. Subtle!?

Now contrast the "imaginary mode" controller with the "perceptual
mode" one:

>I> >O> the controller |I| |O|
  ^ | -------------- ^ |
  p a its world p a
   \-|W'=1|-/ \--|W|-/

The diagram on the right makes sense to me: the controller acts on
the world and its perceives its "reaction", the goal being to control
for ("compute", some say) those actions a that result in the desired
perceptions

  p-desired = r = W * a-appropriate

That is, we have to act appropriately

  a-appropriate = W^-1 * p-desired

to achieve what we want.

But what is the meaning of the diagram on the left? Because W'=1, it
says that the imaginary world has properties (or has it any?) that
have nothing to do with the real world (W). It thus says that

  a-appropriate = p-desired

again without taking into account any of the properties of the real
world, in which the results of the "imagination work" must ultimately
lead to action. Note that I presuppose that imagination somehow
subserves how we will act when _not_ imagining things. If not,
imagination would serve no purpose and could just as well not exist.

Let's forget about words and do some math:

  e = r2 - p2

  p2 = I * O * e

  e = r2 - I * O * e

         r2
  e = ---------
      1 + I * O

Now this cannot be an error in the usual sense: it cannot be
controlled away. The only way e can become zero is through setting
r2=0. Subtle!? Maybe not for a Buddhist who might explain that the
state of being without error is possible only by abolishing all our
desires ;-). Is that what the diagram adstructs?

Another way to make the error small would be through making I * O
large. But, says PCT, that happens anyway. No imagination loop would
be necessary to accomplish this. And in practice making I * O large
_for W'=1_ would often cause instabilities as soon as W reappears and
those values of I and O that were tuned to W'=1 are again used on W,
in the non-imaginary mode. So this is not a promising avenue either.

Finally, what does the loop _accomplish_? Why have a loop that
attempts to bring p2 to r2 if r2 is there all along? Why would we
want p2 if it is superfluous?

What goes on here? I can make no sense of this diagram at all. Help!

My position on the "imagination mode" is this: the imagination mode
is _useful_, i.e. predictive (of the likely perceptual results of
actions) only if the linkage between a and p is not 1 but a W' which
closely resembles the true W.

Greetings,

Hans

[Bill_Powers1]

[From Bill Powers (970918.0926 MDT)]

Hans Blom, 970918c--

Now contrast the "imaginary mode" controller with the "perceptual
mode" one:

>I> >O> the controller |I| |O|
^ | -------------- ^ |
p a its world p a
  \-|W'=1|-/ \--|W|-/

The diagram on the right makes sense to me: the controller acts on
the world and its perceives its "reaction", the goal being to control
for ("compute", some say) those actions a that result in the desired
perceptions

That's because you keep thinking of the imagination connection as
representing a REAL world. For your concept of a behavioral model, that's
necessary. In PCT it is not necessary; the function of imagination is not
to model the world, but only to provide a way of saying "If I _could_
control this perception perfectly, would the result be what I want?"

If you keep trying to interpret the HPCT model in terms of your
single-level model-based control system, you are never going to grasp how
the PCT model works. Since I predict that you will keep doing the former, I
also predict that the latter will continue to be true.

Best,

Bill P.