Reinforcers

[From Bill Powers (2001.01.05.0249 MST)]

The chief reason I'm up at this hour is a thought about trying to define
reinforcements and "establishing conditions."

Here is a proposed definition of a reinforcer:

"If X is a reinforcer to an organism, then any environmental condition or
event tending to change the state of X is followed by an action or change
in action by the organism that tends to restore X to its original state, or
to keep X from changing significantly in the first place."

This proposed definition is supposed to apply not just to the delivery of
positive, negative, or aversive reinforcements through a contingency, but
to the entire process of operant conditioning beginning with setting up the
establishing conditions (that is, the deprivation of X or the noxious
presence of X). It is, of course, a kind of definition of a controlled
variable. I am proposing that X is a reinforcer if and only if it is a
controlled variable.

Setting up the establishing conditions is done by disturbing the state of
X, for example, depriving the animal of food or water, or subjecting it to
repeated shocks. This doesn't tell us _exactly_ what the critical variable
is, but we can reasonably deduce that it is some variable or variables in
the organism which depend on X.

X is discovered to be a reinforcer when it is disturbed from its normal
state and the animal exerts efforts to restore X to its former state. These
efforts might consist of the execution of already-learned actions, or they
might consist of reorganizing the existing behavioral systems until a
behavior pattern results that brings X toward its former state.

I would be interested in hearing of any example of a reinforcer that fails
to meet this definition.

The primary issue between behaviorism and PCT is whether behavior is
controlled by the environment or is produced by the organism as a means of
controlling something in the organism despite changes in the environment.
In considering "establishing conditions" we have to ask why some variables
can be established as reinforcers while others cannot, for a given
organism. Why is it that we can establish food or water deprivation to make
these substances reinforcers for a rat, but not deprivation of neon? The
answer lies in observing the variable prior to setting up the establishing
condition.

What makes an animal eat, or drink, or breath, or seek warmth when the
environment is cold? Does food make the animal eat, does water make it
drink, and so on? No, clearly it is something inside the animal that
demands food, water, air, and warmth. The animal comes to us already
organized to obtain these things and to learn to obtain them when the
environment changes. It is the behavior of the animal, innate or acquired,
that causes the environment to yield up food, water, air, and warmth. And
it is solely the organism that determines that such things are required.

If there is too much X, the animal can reduce it or learn to reduce it. If
there is too little X, the animal can increase it or learn to increase it.
If X is just right, the animal does nothing to change it. The definitions
of too much, too little, and just right are not set by the environment, but
by the organism. Goldilocks, clearly, was a control system.

Best,

Bill P.

[From Chris Cherpas (2001.02.07.1000 PT)]

I agree. Establishers are references (or deviations
from them -- but that just makes it more complicated),
whether way down deep in where the physiological
correlates of food deprivation live, or all the way up
the neural control system hierarchy. The Experimental
Analysis of Behavior never ventured far enough down the
hierarchy to see that "responses" were continuously being
composed, including control all the way down to intensities,
and hence never even considered that there could be
"establishers" that are so subtle with changes and deviations
occurring all the time. The closest to this was Jack
Michael's work, but, of course, he failed. [See if you
agree with what I said in the "Why I stopped reading Verbal
Behavior" article.]

Best regards,
cc

Bill Powers (2001.01.05.0249 MST)--

···

The chief reason I'm up at this hour is a thought about trying to define
reinforcements and "establishing conditions."

Here is a proposed definition of a reinforcer:

"If X is a reinforcer to an organism, then any environmental condition or
event tending to change the state of X is followed by an action or change
in action by the organism that tends to restore X to its original state, or
to keep X from changing significantly in the first place."

This proposed definition is supposed to apply not just to the delivery of
positive, negative, or aversive reinforcements through a contingency, but
to the entire process of operant conditioning beginning with setting up the
establishing conditions (that is, the deprivation of X or the noxious
presence of X). It is, of course, a kind of definition of a controlled
variable. I am proposing that X is a reinforcer if and only if it is a
controlled variable.

Setting up the establishing conditions is done by disturbing the state of
X, for example, depriving the animal of food or water, or subjecting it to
repeated shocks. This doesn't tell us _exactly_ what the critical variable
is, but we can reasonably deduce that it is some variable or variables in
the organism which depend on X.

X is discovered to be a reinforcer when it is disturbed from its normal
state and the animal exerts efforts to restore X to its former state. These
efforts might consist of the execution of already-learned actions, or they
might consist of reorganizing the existing behavioral systems until a
behavior pattern results that brings X toward its former state.

I would be interested in hearing of any example of a reinforcer that fails
to meet this definition.

The primary issue between behaviorism and PCT is whether behavior is
controlled by the environment or is produced by the organism as a means of
controlling something in the organism despite changes in the environment.
In considering "establishing conditions" we have to ask why some variables
can be established as reinforcers while others cannot, for a given
organism. Why is it that we can establish food or water deprivation to make
these substances reinforcers for a rat, but not deprivation of neon? The
answer lies in observing the variable prior to setting up the establishing
condition.

What makes an animal eat, or drink, or breath, or seek warmth when the
environment is cold? Does food make the animal eat, does water make it
drink, and so on? No, clearly it is something inside the animal that
demands food, water, air, and warmth. The animal comes to us already
organized to obtain these things and to learn to obtain them when the
environment changes. It is the behavior of the animal, innate or acquired,
that causes the environment to yield up food, water, air, and warmth. And
it is solely the organism that determines that such things are required.

If there is too much X, the animal can reduce it or learn to reduce it. If
there is too little X, the animal can increase it or learn to increase it.
If X is just right, the animal does nothing to change it. The definitions
of too much, too little, and just right are not set by the environment, but
by the organism. Goldilocks, clearly, was a control system.

[From Bill Powers (2001.01.08.0500 MST)]

Chris Cherpas (2001.02.07.1000 PT)--

I agree. Establishers are references (or deviations
from them -- but that just makes it more complicated),
whether way down deep in where the physiological
correlates of food deprivation live, or all the way up
the neural control system hierarchy. The Experimental
Analysis of Behavior never ventured far enough down the
hierarchy to see that "responses" were continuously being
composed, including control all the way down to intensities,
and hence never even considered that there could be
"establishers" that are so subtle with changes and deviations
occurring all the time. The closest to this was Jack
Michael's work, but, of course, he failed. [See if you
agree with what I said in the "Why I stopped reading Verbal
Behavior" article.]

I'll have to find it, first. Who was Jack Michael, and what did he fail at?
Remember that I'm only an engineer.

It was only a year or two ago that Bruce Abbott mentioned establishing
conditions and it has taken a while for me to recognize that this rounds
out the picture. Reinforcers are controlled variables. If they weren't,
organisms would not do anything to restore them when they are removed, or
the opposite for aversive reinforcers. And it wouldn't be necessary them to
be disturbed (naturally or on purpose) before they could be used as
reinforcers.

Eventually, I suppose, this will sink in and adherents to EAB will
gradually fade away. But I predict that EAB will never die, for the same
reason that there are still flat-earthers and people who believe in the
equivalent of humours. It's my contention that every bad idea ever invented
is still believed by some group of people, somewhere. This is largely
because each new generation has to start from scratch, with education being
only a mildly effective shortcut (as Bruce Gregory and his crowd have
demonstrated). What we now see as incorrect notions were once the pinnacle
of intellectual achievement. They will always seem to make sense, if one
doesn't learn exactly why they were abandoned by most people.

I think I remember hearing, when I was very young, that the phases of the
moon are caused by clouds. As Bruce G.'s Smithsonian group found, some
people still believe that.

Best,

Bill P.

[From Chris Cherpas (2001.02.08.1000 PT)]

The closest to this was Jack
Michael's work, but, of course, he failed. [See if you
agree with what I said in the "Why I stopped reading Verbal
Behavior" article.]

Bill Powers (2001.02.08.0500 MST)--

I'll have to find it, first. Who was Jack Michael,
and what did he fail at?

The article I refer to can be found at:

http://www.ed.uiuc.edu/csg/cherpas1998.html

Jack Michael failed at trying to generalize the
establisher-reinforcer relation, i.e., no reinforcement
without establishment, whether it is "unconditioned"
(like food deprivation) or not. He's referenced in
the article above. If you're really interested, I've
got an article in the Behavior Analyst that goes over
some of the problems (but I'm guessing it won't be that
interesting!).

Eventually, I suppose, this will sink in and adherents to EAB will
gradually fade away. But I predict that EAB will never die, for the same
reason that there are still flat-earthers and people who believe in the
equivalent of humours. It's my contention that every bad idea ever invented
is still believed by some group of people, somewhere. This is largely
because each new generation has to start from scratch, with education being
only a mildly effective shortcut (as Bruce Gregory and his crowd have
demonstrated). What we now see as incorrect notions were once the pinnacle
of intellectual achievement. They will always seem to make sense, if one
doesn't learn exactly why they were abandoned by most people.

I'm way too optimistic by your thinking, then, because
I'd like to see theism descend into a minority position
in my life! Try, try again.

Best regards,
cc

[From Bill Powers (2001.02.09.0704)]

Chris Cherpas (2001.02.08.1000 PT)--

Jack Michael failed at trying to generalize the
establisher-reinforcer relation, i.e., no reinforcement
without establishment, whether it is "unconditioned"
(like food deprivation) or not. He's referenced in
the article above. If you're really interested, I've
got an article in the Behavior Analyst that goes over
some of the problems (but I'm guessing it won't be that
interesting!).

Try me.

I'm glad you reminded us of your article; I had forgotten just how good it
is. I don't suppose there has been any reaction to it from inside EAB.

It seems strange that the relation between establishers and reinforcers was
not clear. The expression "establisher" sounds odd to me -- Bruce A. called
it an "establishing condition" which would seem to make more sense. Being
without the reinforcer for X hours is a condition that can be established,
but as a agency-type noun that gets pretty awkward. The "being without
reinforcement" - er? I suppose that once you get used to saying things a
certain way, you stop listening to what you're saying (although, as you
point out, you still have to _hear_ it).

I haven't seen any direct discussion of these subjects (EAB/PCT) between
you and Bruce Abbott. I'm not trying to start a fight, but it would be
useful to know what agreements and disagreements there are between you two,
since you both come (nominally) from the same field. My impression is that
you and Bruce have arrived at some different conclusions about TEAB, but
perhaps that is just a matter of strategy. Any comments from either of you?

Best,

Bill P.

Many thanks to Bill Powers for the nifty definition of reinforcers. It
goes a long way toward helping me reconcile what I have always thought to
be the "fit" between PCT and the behaviorists' view and it certainly
simplifies matters. I've also pulled down Chris Cherpas' article and will
read it straightaway.

Fred Nickols
The Distance Consulting Company
"Assistance at A Distance"
http://home.att.net/~nickols/distance.htm
nickols@att.net
(609) 490-0095