cooperation and competition

[R_Marken]

[From Rick Marken (2000.05.06.0830)]

Bruce Nevin (2000.0505.1111 EDT) --

If they resolve the conflict so that each is able to
control "happily" (so to speak), they are in cooperation.

Yes. But I think it's important to point out that cooperation
is an active control process, not a happy accident. Cooperative
control makes it possible for independent control systems to
control variables better (or, sometimes, _at all_). Cooperation
is a purposeful (control) activity; it can only occur among
systems that can control for "cooperative" perceptions, such as
"sharing" and "helping"; perceptions which take the behavior
of other control systems into account.

Conflict resolution is one way to cooperate. When there is
conflict, the parties to the conflict, assuming they are of
about equal strength, will not be able to control the variable
in contention (X). They can solve the conflict by agreeing to,
say, control variable X at different times. This solution requires
control of a higher level cooperative perception (perception of
"sharing", say) by both parties. If one or another of the parties
doesn't control for this cooperative perception, then there is
no reason for that party to give up attempts to have complete
control of variable X. A cooperative solution typically involves
each party giving up some control (in this case, giving up the
possibility of complete control of X) in order to achieve even
greater control (the ability to control X at all)..

A more common (and productive) kind of cooperation occurs when
people join together in a way that allows them to control
a variable that they could not control individually. A simple
example occurs when two people coordinate their efforts in order
to lift a couch. The height of the couch off the floor (the
controlled variable) cannot be controlled by either individual
alone. Again, this kind of cooperation cannot occur by accident;
it requires active control of a higher level "cooperative"
perception ("helping", perhaps). In order to control this
cooperative perception, the parties must independently adjust
control of lower level perceptions (of _when_ to exert the lifting
force, for example) so that their efforts are coordinated and
result in the perception not only of a lifted couch but of
having helped each other.

I think it is possible to distinguish this kind of active
cooperative controlling from interactions that _appear_
to be cooperative but are actually just the operation
of independent control systems which are _not_ controlling
for cooperative perceptions. For example, I doubt that the
apparent cooperation between insects, like ants and bees, is
an active control process. Bees, for example, probably dance
to control their own perceptions, not to show other bees where
to find pollen. The dance is probably a disturbance to
perceptions that the worker bees control by flying off in
the direction of the dance.

I think it should be possible to devise a tracking demo that
makes it possible to distinguish an outcome (controlled result)
that is produced by active cooperative control from the same
outcome when it is produced by independent control systems that
are not actively controlling for cooperation. I think those
of you who are interested in cooperation and competition could
learn a lot (and be able to teach us a lot) about cooperative
control if you spent some time devising such a demonstration.

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_Gregory]

[From Bruce Gregory (2000.05.07.1417)]

Rick Marken (2000.05.06.0830)

I think it is possible to distinguish this kind of active
cooperative controlling from interactions that _appear_
to be cooperative but are actually just the operation
of independent control systems which are _not_ controlling
for cooperative perceptions. For example, I doubt that the
apparent cooperation between insects, like ants and bees, is
an active control process. Bees, for example, probably dance
to control their own perceptions, not to show other bees where
to find pollen.

Don't all LCSs act to control their own perceptions? How could bees show
other bees where to find pollen without controlling their own perceptions?

The dance is probably a disturbance to
perceptions that the worker bees control by flying off in
the direction of the dance.

Isn't this true of any action taken "in response" to another's actions?

I think it should be possible to devise a tracking demo that
makes it possible to distinguish an outcome (controlled result)
that is produced by active cooperative control from the same
outcome when it is produced by independent control systems that
are not actively controlling for cooperation. I think those
of you who are interested in cooperation and competition could
learn a lot (and be able to teach us a lot) about cooperative
control if you spent some time devising such a demonstration.

Driving in traffic might be a good example. I suspect that few drivers are
actively controlling for cooperation.

BG

[R_Marken]

[From Rick Marken (2000.05.07.1720)]

Me:

Bees, for example, probably dance to control their own
perceptions, not to show other bees where to find pollen.

Bruce Gregory (2000.05.07.1417)--

Don't all LCSs act to control their own perceptions?

Yes. Of course. I was trying to say that cooperation --
real cooperative controlling, as in the "carrying the
couch" example -- requires that people intentionally
agree to control certain perceptions (like "lift couch on
signal") in order to produce a particular result. That
is, cooperation requires that people control "cooperative
perceptions", like "sharing", "helping", etc.

How could bees show other bees where to find pollen
without controlling their own perceptions?

My point was that bees can _appear_ to be cooperating without
actually _intending_ to cooperate. I think the bees produce their
"cooperative results" (like building hives) in the same way
that a subject achieves the "cooperative result" of controlling
a cursor in two dimensions (see the "Degrees of Freedom" chapter
in _Mind Readings_). Control actions in one dimension can be
made to produce disturbances in another; actions that compensate
for these disturbance can also cause disturbances in the first
dimension. But two _independently acting_ control systems
_coordinate_ their actions and compensate for these disturbances,
producing the _appearance_ of "cooperation". It looks like the
two independent control systems are cooperating with each
other to generate the actions that produce the result; but, in
fact, the two systems know nothing about each other. The apparent
"cooperation" is just the simultaneous controlling of two
completely independent control systems.

I think what is going on in the two dimensinoal tracking task
is exactly what is going on with bees -- except that the
independent control systems are in seperate organisms rather
than in the same organism. The bees end up takng all the actions
(dancing, getting pollen, etc) that end up producing a hive,
I think, for basically the same reason that the subject in
the two-dimensional tracking task ends up taking all the
actions that end up keeping the cursor on target; it's a
result that "falls out" of the operation of independent
control systems.

In the case of the bees, these control systems are inherited;
some bees inherit systems that are designed to keep certain
kinds of perceptions at certain references; others inherit
systems that are designed to keep other kinds of perceptions
at other references. Different kinds of bees (workers, seekers,
queens, etc) inherit the ability (and need) to control
different kinds of perceptual variables; and they control
them as best as they can. The result is the _appearance_ of
complex cooperative control. I doubt (though this should be
tested by PCT aware naturalists) that there is any cooperation
among bees -- real, active cooperation of the kind that occurs
when humans cooperate to build a skyscraper or a computer chip.

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_Gregory]

[From Bruce Gregory (2000.05.08.0956)]

Rick Marken (2000.05.07.1720)

Yes. Of course. I was trying to say that cooperation --
real cooperative controlling, as in the "carrying the
couch" example -- requires that people intentionally
agree to control certain perceptions (like "lift couch on
signal") in order to produce a particular result. That
is, cooperation requires that people control "cooperative
perceptions", like "sharing", "helping", etc.

O.K. Although it is not obvious to me that one cannot carry a couch
without controlling "cooperative perceptions." I can't imagine that it
is not possible to carry a couch from one room to another without
invoking a model that requires such high order perceptions. Probably
just my lack of imagination.

My point was that bees can _appear_ to be cooperating without
actually _intending_ to cooperate.

I would be very surprised if the actions of any animal (other than
possibly a human animal)could not be modeled without invoking a
controlled perception at the level of "intending to cooperate".
Therefore cooperation as understood in PCT seems to be purely a human
phenomenon. Furthermore, many examples that would be labeled cooperation
(such as alternate merges) are not "really" cooperation since there may
be no controlled perception of "cooperation".

I think the bees produce their
"cooperative results" (like building hives) in the same way
that a subject achieves the "cooperative result" of controlling
a cursor in two dimensions (see the "Degrees of Freedom" chapter
in _Mind Readings_). Control actions in one dimension can be
made to produce disturbances in another; actions that compensate
for these disturbance can also cause disturbances in the first
dimension. But two _independently acting_ control systems
_coordinate_ their actions and compensate for these disturbances,
producing the _appearance_ of "cooperation". It looks like the
two independent control systems are cooperating with each
other to generate the actions that produce the result; but, in
fact, the two systems know nothing about each other. The apparent
"cooperation" is just the simultaneous controlling of two
completely independent control systems.

O.K. So in PCT cooperation requires that one be aware of the intentions
of others and be controlling for the perception "cooperating".

I think what is going on in the two dimensional tracking task
is exactly what is going on with bees -- except that the
independent control systems are in separate organisms rather
than in the same organism. The bees end up taking all the actions
(dancing, getting pollen, etc) that end up producing a hive,
I think, for basically the same reason that the subject in
the two-dimensional tracking task ends up taking all the
actions that end up keeping the cursor on target; it's a
result that "falls out" of the operation of independent
control systems.

Doubtless this is true of much, if not all, of the behavior we
PCT-mislabel as cooperative.

In the case of the bees, these control systems are inherited;
some bees inherit systems that are designed to keep certain
kinds of perceptions at certain references; others inherit
systems that are designed to keep other kinds of perceptions
at other references. Different kinds of bees (workers, seekers,
queens, etc) inherit the ability (and need) to control
different kinds of perceptual variables; and they control
them as best as they can. The result is the _appearance_ of
complex cooperative control. I doubt (though this should be
tested by PCT aware naturalists) that there is any cooperation
among bees -- real, active cooperation of the kind that occurs
when humans cooperate to build a skyscraper or a computer chip.

It's not obvious to me that even in these enterprises (building
skyscrapers or computer chips) people are cooperating in the PCT sense
much of the time.

BG

[R_Marken]

[From Rick Marken (2000.05.08.1050)]

Bruce Gregory (2000.05.08.0956)--

O.K. Although it is not obvious to me that one cannot
carry a couch without controlling "cooperative perceptions."

Think of it this way: In order to control for carrying the
couch both people have to adopt this goal ("couch lifted")
_at the same time_. I have to grasp and lift one end of the
couch when you grasp and lift the other. People don't have
built in references for going over and lifting one end of a
couch when someone goes over and lifts the other. They have
to communicate with each other and agree to adopt the goal
of lifting the couch, then agree to lift opposite ends at
the same time. All this "agreeing" is what I would call
cooperative control; it will only happen to the extent that
people are controlling for perceptions like "helpfulness",
"friendship" or whatever.

Bees don't have to control for these cooperative perceptions
in order to behave in a coordinated manner. Bees have genetically
built in references that lead to coordinated action. When a bee
sees another bee doing the equivalent of trying to lift one end
of a couch, the bee goes right over and starts lifting the other
end because bees _do_ have built in references for perceiving the
equivalent of a lifted coach.

Another way to look at it is like this: In bees, cooperation is
not a choice. Because the bees have genetic references for seeing
the equivalent of a lifted couch, they will add their efforts to
any existing bee efforts in order to produce this perception for
themselves. In humans, cooperation is a choice. Because people
have no genetic reference for seeing a lifted couch, they will
add their efforts to their friend's efforts only if they are
controlling for cooperative (probably system concept) perceptions;
if they are not controlling for such perceptions, they will sit
back and control for whatever other perceptions they care about
while their friend struggles with the couch.

Best

Rick

···

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

[bnhpct]

[From Bruce Nevin (2000.05.08 1928 EDT)]

Rick Marken (2000.05.07.1720)--

···

At 05:23 PM 05/07/2000 -0800, Rick Marken wrote:

Bees ... probably [don't] dance to show other bees where to find pollen.

How would you Test this hypothesis?

        Bruce Nevin

[Bruce_Gregory]

[From Bruce Gregory (2000.05.09.0952)]

Rick Marken (2000.05.08.1050)

Another way to look at it is like this: In bees, cooperation is
not a choice. Because the bees have genetic references for seeing
the equivalent of a lifted couch, they will add their efforts to
any existing bee efforts in order to produce this perception for
themselves. In humans, cooperation is a choice. Because people
have no genetic reference for seeing a lifted couch, they will
add their efforts to their friend's efforts only if they are
controlling for cooperative (probably system concept) perceptions;
if they are not controlling for such perceptions, they will sit
back and control for whatever other perceptions they care about
while their friend struggles with the couch.

If I read you correctly, you are saying that modeling "cooperation"
between human beings requires a more complex hierarchy than does
modeling "cooperation" between bees. I agree. (Choice is not defined in
HPCT, at least as far as I know, so I would prefer to avoid the term
since it is so easy to misinterpret.)

BG

[Bruce_Gregory]

[From Bruce Gregory (2000.05.09.0945)]

Bruce Nevin (2000.05.08 1928 EDT)

Rick Marken (2000.05.07.1720)--

>> Bees ... probably [don't] dance to show other bees where
to find pollen.

How would you Test this hypothesis?

Presumably Rick is maintaining that bees are not controlling a higher
order perception that might be labeled "showing other bees where to find
pollen." If the first bee is prevented from dancing, it presumably will
not attempt to communicate the location of the pollen using some other
"symbol".

BG

···

At 05:23 PM 05/07/2000 -0800, Rick Marken wrote:

[bnhpct]

[From Bruce Nevin (2000.05.09 1928 EDT)]

Bruce Gregory (2000.05.09.0945)

···

At 09:46 AM 05/09/2000 -0400, Bruce Gregory wrote:

If the first bee is prevented from dancing, it presumably will
not attempt to communicate the location of the pollen using some other
"symbol".

This disturbance overpowers the bee's ability to control. Is that
appropriate for the Test for a controlled variable? It also assumes that
the bee has other means at its disposal for communicating the location of
something that it perceives that the hive needs.

        Bruce Nevin

[R_Marken]

[From Rick Marken (2000.05.09.0800)]

Me:

Bees ... probably [don't] dance to show other bees where
to find pollen.

Bruce Nevin (2000.05.08 1928 EDT) --

How would you Test this hypothesis?

The same way you test any hypothesis about a controlled
variable. The first step is to define the hypothetical
controlled variable as precisely as possible. In this case,
you might start by saying that when a bee "dances" it is
controlling a perception of its direction of movement relative
to, say, the sun. This definition makes the presence of other
bees irrelevant. So you can test this hypothesis by removing
bees from the hive when the dancer returns. If the dance goes
on anyway, that is one little piece of evidence that the
hypothesis about the controlled variable is correct (or, at
least, not wrong). Also, changing the apparent location of
the sun using an artificial light source would be another test.
If the orientation of the dance changes in proportion to the
change in the apparent location of the sun then this is another
piece of evidence that the hypothesis about the controlled
variable is correct. Of course, if any of these tests fails,
then the hypothesis about the controlled variable must be
revised and tested further.

I think that many ethological studies embody various aspects
of the test for the controlled variable. What usually seems
to be missing from these studies is 1) a firm grasp of what
controlled variables _are_ 2) an understanding that organisms
control many variables _simultaneously_ 3) a grasp of the fact
that a controlled variables must be defined from the point of
view of the behaving organism, not from the point of view of
the observer (this is why the observation of "open loop" egg
rolling behavior in herring gulls is so far off the mark) and
4) an understanding of the fact that testing must be an iterative
process that involves continual revision of hypotheses about
the controlled variable based on the results of testing. If
just one of these aspects of the test is missing, then its
really impossible to be confident that anything close to an
actual controlled variable has been identified.

Bruce Gregory (2000.05.09.0952)--

If I read you correctly, you are saying that modeling
"cooperation" between human beings requires a more complex
hierarchy than does modeling "cooperation" between bees.
I agree.

I think what I've been trying to say is that there are really two
kinds of "cooperation"; _passive_ cooperation, where coordinated
activities result from genetic or higher level systems setting
references for two or more systems in such a way that the systems
act "cooperatively" to produce a particular result (this is the
kind of cooperation we see in the 2-D tracking task and the behavior
of social insects) and _active_ cooperation, where independent
systems have to "agree" to set their reference for perceptions
in such a way that they act "cooperatively" to produce a
particular, mutually desired result. I really can't think of
many tests to determine which kind of cooperation is occurring in
a particular situation other than by looking to see whether of not
some kind of agreement process took place prior to performance
of the cooperative activity. Of course, this view of cooperation
suggests that active cooperation can only occur among organisms
that can communicate in some way (apes, hominids, porpoises?).

Best

Rick

···

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

[Bruce_Gregory]

[From Bruce Gregory (2000.05.09.1132)]

Bruce Nevin (2000.05.09 1928 EDT)

This disturbance overpowers the bee's ability to control. Is that
appropriate for the Test for a controlled variable?

It depends on what you think the controlled variable is. If I call you
and get a busy signal, is this disturbance overpowering my ability to
control? Not if I send you this e-mail message.

It also
assumes that
the bee has other means at its disposal for communicating the
location of
something that it perceives that the hive needs.

Indeed.

BG

[bnhpct]

[From Bruce Nevin (2000.05.09 1147 EDT)]

Bruce Gregory (2000.05.09.1132)--

If I call you
and get a busy signal, is this disturbance overpowering my ability to
control? Not if I send you this e-mail message.

Comparable to preventing the bee from dancing would be some disturbance
that prevented you from using the means that are established (in our shared
culture) for communicating whatever it is that you want to tell me. No
language, no conventional gestures.

I expect that the bee would resist whatever was preventing it from dancing
its news. Whether it has other means for delivering the news about location
and direction depends upon the other bees as much as it does on the one
that you are preventing from dancing.

If it gave up, delivered its load (if it had nectar or water), took off
again for the source that it had found (bees can and frequently do follow a
forager that is moving "in a beeline" rather than in more or less random
search), and on return attempted to dance the location again, would that count?

        Bruce Nevin

···

At 11:34 AM 05/09/2000 -0400, Bruce Gregory wrote:

[bnhpct]

[From Bruce Nevin (2000.05.09 1200 EDT)]

Rick Marken (series of posts of recent date)--

Your view, as I understand it, is that control systems may appear to be
cooperating without cooperation being intended, and that we should limit
the use of the word "cooperation" to those cases where cooperation is intended.

In addition, we should not use the word (you say) when the references for
cooperation are inherited, as you believe is the case with bees.

If we found freeloaders among bees, we could hypothesize that the genetic
"programming" failed in those individuals. If we could induce freeloading
by changing the body chemistry (not the genetics) of bees, we could say
that these changes in body chemistry interfered with the genetic
"programming". We might find a genetic susceptibility to body chemistry
being interfered with by certain circumstances during maturation.

There is some evidence for a genetic predisposition among mammals for
maintaining alliances. Counterexamples might be explained as the lack of
this "programming" in some individuals, or by finding that it is dampened
or thwarted by changes in body chemistry, related to certain circumstances
during maturation, or a combination of both. (The example of the two
contrasted kangaroo mothers comes to mind, and investigations into the
neurochemistry of psychopaths.) Would you not then have to avoid use of the
word "cooperation" when cooperation appears to be going on?

Consider now non-biological inheritance, culture in the sense of ethnology
and anthropology. In some cultures, it is difficult or impossible to
imagine non-cooperative action, even when an individual is not in company
of others--that is, to imagine not coordinating one's purposes with those
of others, with the aim of benefit (which is understood to be necessarily
mutual) and the avoidance of harm (which is also understood to be
necessarily mutual). Put another way, if harm is intended, it is understood
to be necessarily mutual in such a culture. In such a "high synergy"
culture, would you not have to avoid use of the word "cooperation," because
the people have no choice?

There is the further matter of communicating and agreeing to adopt a
particular goal.

Rick Marken (2000.05.08.1050)--

[People] have to communicate with each other and agree to
adopt the goal of lifting the couch, then agree to lift
opposite ends at the same time. All this "agreeing" is what
I would call cooperative control; it will only happen to the
extent that people are controlling for perceptions like
"helpfulness", "friendship" or whatever.

This concerns a process of each setting references the same as the other is
setting them--couch in moving van, for example. This is a prerequisite for
cooperation. But it also concerns ongoing control of interfering with the
other's control. "Not too fast, my hand is slipping."

Among bees, the goals are pre-established: maintain certain necessities of
life in a hive near optimum levels: nectar, water, temperature, size of
hive, density of population (size relative to current hive space), perhaps
some others. The dance of a forager may be about any of these things. The
dancer smells of nectar, or water, or damp wood as of a hollow in a tree
where a new hive might lodge. If a hive-mate perceives a need for what they
smell, she goes foraging in the direction and at the approximate distance
indicated by the dance; those that do not perceive a need for what they
smell on the dancer ignore the dancer.

The agreement about maintaining optimum levels of what bees in a hive need
is not established at the time of the dance.

The fact that helping someone is significant for maintaining an alliance,
and that withholding help generally demonstrates non-alliance, is an
"agreement" that is not established at the time of giving help.

        Bruce Nevin

···

At 11:00 AM 05/08/2000 -0700, Richard S. Marken wrote:

[bnhpct]

[From Bruce Nevin (2000.05.09 1342 EDT)]

Rick Marken (2000.05.09.0800)

The first step is to define the hypothetical
controlled variable as precisely as possible. In this case,
you might start by saying that when a bee "dances" it is
controlling a perception of its direction of movement relative
to, say, the sun. This definition makes the presence of other
bees irrelevant. So you can test this hypothesis by removing
bees from the hive when the dancer returns.

If the bee failed to dance it could be because absence of other bees
disturbs other perceptions that the bee is controlling. In particular, if
the bee is controlling some perception that we could label "communicating"
then absence of any bees to communicate its news to would surely disturb
its control of that perception.

One group used little robot bees. However, the mechanical bees tended to
bulldoze through the real bees in the hive, not interacting with them as a
live bee did, and perhaps this is one reason why the real bees ignored
them. Bearing in mind that a real bee is ignored by her hive-mates if they
don't perceive a need for whatever it is she is telling them about (as
determined, evidently, by smell).

changing the apparent location of
the sun using an artificial light source would be another test.
If the orientation of the dance changes in proportion to the
change in the apparent location of the sun then this is another
piece of evidence that the hypothesis about the controlled
variable is correct.

I think this has been done. In any case, the diurnal movement of the sun
appears to be a disturbance that the bees resist by altering the angle of
the straight (overlapping) portion of the figure-8 dance relative to gravity.

I will be away the rest of today.

        Bruce Nevin

···

At 08:00 AM 05/09/2000 -0700, Richard S. Marken wrote:

[R_Marken]

[From Rick Marken (2000.05.09.1100)]

Bruce Nevin (2000.05.09 1200 EDT)]

Your view, as I understand it, is that control systems may
appear to be cooperating without cooperation being intended,
and that we should limit the use of the word "cooperation" to
those cases where cooperation is intended.

Yes. But I don't really think it's necessary to limit use of
the word "cooperation" to those cases where cooperation is
intended. I think it's good enough to distinguish _intended_
(I called it "active" in the previous post) cooperation from
_unintended_ (I called it "passive" in the previous post)
cooperation. In the first case, cooperation is a controlled
result; in the second, it's not. If you can define what
"cooperation" is in terms of controlled variables, then
you can test to see whether cooperation is intended or not.

Best

Rick

···

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

[Bruce_Gregory]

[From Bruce Gregory (2000.05.10.1118)]

Rick Marken (2000.05.09.1100)

Yes. But I don't really think it's necessary to limit use of
the word "cooperation" to those cases where cooperation is
intended. I think it's good enough to distinguish _intended_
(I called it "active" in the previous post) cooperation from
_unintended_ (I called it "passive" in the previous post)
cooperation. In the first case, cooperation is a controlled
result; in the second, it's not. If you can define what
"cooperation" is in terms of controlled variables, then
you can test to see whether cooperation is intended or not.

Sounds perfectly reasonable to me.

BG

[R_Marken]

[From Rick Marken (2000.05.10.0850)]

Me:

The first step is to define the hypothetical
controlled variable as precisely as possible. In this case,
you might start by saying that when a bee "dances" it is
controlling a perception of its direction of movement relative
to, say, the sun. This definition makes the presence of other
bees irrelevant. So you can test this hypothesis by removing
bees from the hive when the dancer returns.

Bruce Nevin (2000.05.09 1342 EDT)--

If the bee failed to dance it could be because absence of
other bees disturbs other perceptions that the bee is controlling.

The Test is not a one shot process. If ethologists understood
the Test, they would apply many different disturbances to a
hypothetical controlled variable. Only if _all_ of these
disturbances, judiciously applied (not too abrupt or large),
failed to have any effect on the controlled variable, would
the ethologist conclude that the variable is actually under
control.

As you know, PCT implies a whole new approach to behavioral
research, one aimed at determining controlled variables. Since
behavioral researchers don't know what a controlled variable
_is_, they have never really done research aimed at determining
controlled variables. The results of current ethological
research just gives some vague hints and suggestions about
what variables might be under control. Knowledge of controlled
variables will come only when ethologists learn what "control
of perception" means.

Best

Rick

···

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

[bnhpct]

[From Bruce Nevin (2000.05.10 1310 EDT)]

Rick Marken (2000.05.10.0850)--

The first step is to define the hypothetical
controlled variable as precisely as possible. In this case,
you might start by saying that when a bee "dances" it is
controlling a perception of its direction of movement relative
to, say, the sun. This definition makes the presence of other
bees irrelevant. So you can test this hypothesis by removing
bees from the hive when the dancer returns.

The Test is not a one shot process. If ethologists understood
the Test, they would apply many different disturbances to a
hypothetical controlled variable. Only if _all_ of these
disturbances, judiciously applied (not too abrupt or large),
failed to have any effect on the controlled variable, would
the ethologist conclude that the variable is actually under
control.

I appreciate the aptness of your critique of research without understanding
of what a controlled variable is. My question comes from genuine puzzlement
about how to isolate a single variable in a domain of naturalistic
observation like this. Removing all other bees from the hive is a
disturbance that seems to me to be both abrupt and large, and a disturbance
to a great many likely variables at once.

        Bruce Nevin

···

At 08:50 AM 05/10/2000 -0700, Richard S. Marken wrote:

[BJORN_SIMONSEN]

from [Bjoern Simonsen (2000.05.10.22:40 GMT+1)]

[From Rick Marken (2000.05.09.1100)]

Yes. But I don't really think it's necessary to limit use of
the word "cooperation" to those cases where cooperation is
intended. I think it's good enough to distinguish _intended_
(I called it "active" in the previous post) cooperation from
_unintended_ (I called it "passive" in the previous post)
cooperation. In the first case, cooperation is a controlled
result; in the second, it's not. If you can define what
"cooperation" is in terms of controlled variables, then
you can test to see whether cooperation is intended or not.

You use the concept "intended" cooperation as apposed to "unintended"
cooperation and I am not quite sure if I understand what you mean.

Do I misunderstand you if _intended cooperation_ is _purposeful
cooperation_?

Mary Powers once argued that all behavior is purposeful. Do you agree?

If you agree, what is then _unintended cooperation_ (purposeless
cooperation)?

When you use the words _active cooperation_ and _intended cooperation_ I am
near to think about
"coercion". (In my PCT dictionary I find one of your sentences:
I use the word "coercion" to describe the control of the behavior
of a weaker by a stronger control system.)

I will say that two control systems cooperate if they control their
perceptions at the same time they disturb each other.

Bjoern

[R_Marken]

[From Rick Marken (2000.05.10.1500)]

Bruce Nevin (2000.05.10 1310 EDT)--

I appreciate the aptness of your critique of research
without understanding of what a controlled variable is. My
question comes from genuine puzzlement about how to isolate
a single variable in a domain of naturalistic observation
like this.

There is really no problem. Organisms control thousands
of variables simultaneously but this is not a factor in
doing the Test because the Test is based on observing the
behavior of possible controlled variables, not on the
actions that presumably keep these variables under control.

Suppose, for example, that a person is controlling two
variables simultaneously: x+y and x-y. Any disturbance
to x+y is also a disturbance to x-y. So the action you
take to correct the disturbance to x+y depends on how
this action influences x-y. But this doesn't matter to
the Test. If x+y is under control, then disturbances
to x+y will have little or no effect on the value of
x+y, regardless of how these disturbances affect x-y
and the actions required to keep x-y under control.
Regardless of what other variables an organism is
controlling, if x+y is under control then you can tell
that this is the case by applying disturbances that
would _change_ the value of x+y if x+y were _not_ under
control and observing that x+y is basically unaffected
by the disturbance.

Removing all other bees from the hive is a
disturbance that seems to me to be both abrupt and large,
and a disturbance to a great many likely variables at once.

If this disturbance has an effect on a variable that you
really think is controlled, then you might want to try
applying a gentler disturbance to see if its effect on the
hypothetical controlled variable resulted from the fact
that this disturbance was too abrupt and large rather than
from the fact that the hypothetical controlled variable
was _not_ really under control.

I hope you can see, now, that it is not necessary to avoid
disturbances that may have effects on a great many other,
likely controlled variables. To do the Test you just have
to apply disturbances to a variable you think is controlled
and watch for _lack of effect_ of the disturbance. The central
feature of the Test is monitoring the state of the hypothetical
controlled variable, _not_ the state of the outputs that are
presumably used to protect that variable from disturbance.
That's why conventional research is nearly always useless;
conventional researchers are watching (at best) disturbance-
output relationships, which don't necessarily tell you
anything about a variables that is actually under control.

Best

Rick

···

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