universal error curve, Evol. Psych. and PCT

CSGnet Archive CSG1998
1

[From Rick Marken (980613.1245)]

Chris Cherpas (980612.1600 PT) --

The topic of "giving up" reminds me of the EAB research on
progressive ratio schedules: each time the rat completes a
ratio, the ratio requirement is increased, until finally, the rat
reaches a breaking point where no bar-presses are observed for a
substantial period of time (e.g., 10 minutes). The rat is still
hungry, and each bar-press would bring it one step closer to food,
but it stops pressing. Does this situation describe anything
related to what you are discussing?

Yes. Definitely. If the rat is controlling some aspect of the rein-
forcement (and there is still some question about whether rats can
actually control reinforcement _at all_ in the typical operant
situation) then what you describe looks like "giving up" to me.

Another example of the phneomenon of "giving up" that comes to
mind occurs in studies of "learned helplessness". I was too
squeemish to read these studies in detail (I'm a bleeding heart
liberal when it comes to the treatment of "lower" animals, too)
but I think they would throw rats into canisters of water with
slick sides that could not be scaled. The rats would struggle for
a while but eventually "give up" and just drown. (Some of the work
was done by another Bruce named Overmeyer at the U of M. I knew
Bruce and he was a real nice guy -- though the rats probably
didn't think so;-))

Bob C.(980613.0048) --

First, I think taking into account evolution will help constrain
the search for variables people and animals control especially
when what they are actually controlling for is different than what
they are purportedly controlling.

I think this is a good point. But I think it will work both ways.
Evolutionary considerations may constrain our _hypotheses_ about
the kids of variables an organism is controlling. But then we have
to test these hypotheses (using the Test for the Controlled Variable).
The Test may reveal that our hypotheses were wrong; that the organism
is controlling variables that are quite different than the ones
suggested by evolutionary considerations. This should lead to a
reevaluation of the evolutionary considerations that led us to our
initial hypotheses about the variables the organism controls.

Second, I agree wtih others who think there is a blind variation
and selective retention process within an individual and that this
is a big piece of how reorganization works.

Control theory suggests an interesting alternative to the "natural
selection" model of evolution; I'd call it "purposeful selection".
In natural selection, the selecting is done by the environment;
variants that survive and reproduce are selected. In purposeful
selection, the selecting is done by the organisms themselves;
variants that allow good control are selected because they reproduce
sans mutation; variants that allow only poor control are not
selected because they are more likely to reproduce with (not
necessarily "good") mutation.

I must caution that my primary mission in life at this point is
to be the best clinical psychologist I can be.

That sounds like a very good goal. I think PCT can defintely help
you achieve this goal. And I think you can also help PCT as you
go about trying to achieve this goal.

Best

Rick

···

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

2

[From Bruce Abbott (980614.0750 EST)]

Rick Marken (980613.1245)]

Chris Cherpas (980612.1600 PT)

The topic of "giving up" reminds me of the EAB research on
progressive ratio schedules: each time the rat completes a
ratio, the ratio requirement is increased, until finally, the rat
reaches a breaking point where no bar-presses are observed for a
substantial period of time (e.g., 10 minutes). The rat is still
hungry, and each bar-press would bring it one step closer to food,
but it stops pressing. Does this situation describe anything
related to what you are discussing?

Yes. Definitely. If the rat is controlling some aspect of the rein-
forcement (and there is still some question about whether rats can
actually control reinforcement _at all_ in the typical operant
situation) then what you describe looks like "giving up" to me.

There is still some question about whether rats can actually control
reinforcement _at all_ in the typical operant situation? Hmmm. I wonder
how they manage to get that pellet to fall into the cup, then. . . .

The phenomenon Chris refers to is called "ratio strain." It is indeed a
species of "giving up" and thus might qualify for the "universal error
curve" explanation, but there is good reason to believe that it has another
cause. If the animal receives _all_ its food by earning it in the chamber
(rather than receiving supplemental feeding in the home cage after the
experimental session), ratio strain disappears.

Another example of the phneomenon of "giving up" that comes to
mind occurs in studies of "learned helplessness". I was too
squeemish to read these studies in detail (I'm a bleeding heart
liberal when it comes to the treatment of "lower" animals, too)
but I think they would throw rats into canisters of water with
slick sides that could not be scaled. The rats would struggle for
a while but eventually "give up" and just drown. (Some of the work
was done by another Bruce named Overmeyer at the U of M. I knew
Bruce and he was a real nice guy -- though the rats probably
didn't think so;-))

Well, if you _had_ actually read those studies, you would have realized that
they were not the one you described, which was not a "learned helplessness"
experiment. The experiment you described was performed by Curt Richter
many, many years ago and involved wild rats whose whiskers had been trimmed
(for some purpose relevant to the original intention of the experiment).
Contrary to popular belief, rats are very good swimmers, but when these rats
were placed in a water tank, they swam only briefly, then sank to the
bottom. Although they were rescued immediately, they could not be revived.
Richter was able to show that this effect occurred only in wild rats, and
only in wild rats that had had their whiskers trimmed. The trimming
required that the rats be held tightly, and this evidently was an extremely
stressful experience for the wild rats. When they were then placed in the
tank immediately after the trim, they suffered a severe "parasympathetic
rebound," which stopped their hearts.

Bruce Overmier did indeed study the "learned helplessness" phenomenon, along
with Martin Seligman and Stephen Maier. These experiments involved giving
dogs pre-exposure to inescapable shock, escapable shock, or no shock, and
then placing the dogs in a different environment (a shuttlebox) where they
could easily escape or avoid all shocks by jumping over a hurdle each time a
tone sounded. Dogs receiving no preshock or escapable preshock readily
learned the avoidance task, but those receiving inescapable preshock had
great difficulty making the connection between jumping over the hurdle and
shock termination and avoidance. From the PCT perspective, it is a case of
failing to reorganize so as to develop an effective shock control system.

Overmier, Seligman, and Maier were grad students in Richard Solomon's lab at
Penn State when this research was done, in the early 70s.

Regards,

Bruce

3

[from Tracy Harms (980614.1200)]

Rick Marken (980613.1245)

Control theory suggests an interesting alternative to the "natural
selection" model of evolution; I'd call it "purposeful selection".
In natural selection, the selecting is done by the environment;
variants that survive and reproduce are selected. In purposeful
selection, the selecting is done by the organisms themselves;
variants that allow good control are selected because they reproduce
sans mutation; variants that allow only poor control are not
selected because they are more likely to reproduce with (not
necessarily "good") mutation.

I think I've said this before, but this "alternative" does not look like
any alternative whatsoever. It entails no modification of orthodox
neo-darwinian theory. True, evolutionary theorists have been a little slow
at giving adequate attention to preferential components, but it is
definitely something I run into as I read mainstream evolutionary theory.
PCT has something to add to the discussion; of this I have no doubt. But
so far I just don't see PCT giving rise to a significant alternative to
selectionist theory.

T. Harms

4

[From Rick Marken (980614.1545)]

Tracy Harms (980614.1200)

I think I've said this before, but this "alternative" does not
look like any alternative whatsoever. It entails no modification
of orthodox neo-darwinian theory.

I have not seen evolution modeled as a control process, other than
by Bill Powers. Richard Dawkins (who I think is considered an expert)
clearly places selection in the environment; this is the point
of the "evolution game" he describes in one of his books. In that
game the player (environment) controls the evolution of the
pseudo "organism". If you have a reference to a "neo-darwinian"
model of evolution that is equivalent to the PCT "purposeful
evolution" model I would sure like to see it.

Me:

Rats _cause_ the pellets to fall by pressing the bar..they don't
_control_ anything about the pellets.

Bruce Abbott (980614.1450 EST) --

Well Rick, let me see: If there is no pellet available in
the chamber, the rat presses the lever to obtain one. When
the rat consumes the pellet (the result of the action of
another control system), this disturbs the variable "pellet
availability" from its reference value of "available." To
correct the error, the rat presses the lever, which restores
the variable to the "pellet present" reference state. Sounds
like the rat is controlling something about the pellet (its
availability) to me.

This is not the way you test for controlled variables. All you
are doing is giving a control theory spin to a description of
what you see the rat doing. I could give an S-R spin to the
same description: "no pellet" causes a bar press which causes a
pellet which causes eating which causes "no pellet" which...
Sounds to me like the rat is being caused to behave by the pellet;-)

The correct way to test to determine whether a variable is under
control is by doing (you got it) The Test for the Controlled
Variable -- a Test that has never been performed by students of
"operant conditioning". From the data we do have, it appears that
reinforcement is _not_ controlled because rats do not vary their
output to compensate for changes in the feedback function
("schedule").

Wow, me arguing that it's control and you arguing that it's S-R.
Now _there's_ irony for you.

No irony at all. I only "argue for control" when there is overwhelming
evidence that there _is_ a variable under control. Recall that the null
hypothesis of the Test is that there is no control; that the behavior
we see is a result of ordinary open loop processes. Since you are
not willing to study organisms using the methods of control theory,
you have no way of telling (other than by your personal interpretation)
_whether_ control is going on and (if, in your opinion, it is going
on) _what_ variable is under control.

Me:

Ah. The "trim" _caused_ a "parasympathetic rebound" that _caused_
a heart attack. Nice to see that our PCT man in Indiana is still
on the job;-)

Bruce:

As you know, Rick, disturbances to a controlled variable constitute
one of two truly independent variables or "causes" acting on a
standard PCT-type control system.

Acting on a _controlled variable_, not on a _control system_.

So you see, Rick, I'm not talking lineal cause-and-effect here, but
the behavior (and failure) of a complex control system under strong
and rapidly changing disturbance.

Yes, Bruce. You really know how to talk about control systems. If
you only knew how to _study_ them.

Regards,

Bruce

···

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