[From Bill P. 950710.1815]
Hi, folks --
I've been out of the loop all day. Mary had a nose job (sinuses and
deviated septum, not external) to see if that would put an end to the
cold-bronchitis-asthma sequence. One-day surgery; she's home already and
feeling unexpectedly good. We'll see how this theory works out.
···
-----------------------------------------------------------------------
Bruce Abbott (950710.1035 EST) --
As you note, Skinner labored mightily to avoid terms like
"intentions," "goals, "plans," and the like. It was not that he
believed that purposive behavior does not exist, but rather that he
believed that purposive behavior had been accounted for by the
mechanism of selection-by-consequences and that the use of these
terms tended to lead one to view goals, intentions, and the like as
causes of behavior when in fact those causes resided outside the
organism, in the environment. These words smacked to Skinner of
mentalism, properties of the inner ghost.
Yes, and to a lot of other psychologists, biologists, neurologists,
etc., as well. Your allusion to "ghosts" is particularly appropriate,
because most people who admire themselves for being scientists see any
alternative to external control of behavior as necessarily resorting to
mysticism. I'd say it is basically their ignorance of modern technology
that leads them to this conclusion. Arthur C. Clark again: the products
of any sufficiently advanced civilization will seem to us to work by
magic. In this case, modern technology looks like magic to the old-time
believers in external causality. Skinner assumed that internally
generated behavior had to be random or "capricious." But that is only
because he knew nothing of electronic systems. He had a Tom Swift
acquaintance with technology, not an engineer's. He lived in a world
where scientific instruments came in walnut cases.
Also, your "when in fact those causes resided outside the organism" is a
particularly apt way of stating the matter from the point of view of the
guy who can see the Necker cube in only one way. To Skinner this was a
self-evident fact, not even worth arguing about -- or thinking about.
-------------------------------
If you happen to be unfortunate enough to live in a society where
the history of reinforcement has led to the frequent use of whips
and chains on its citizens, too bad for you, but it's just an
accident of history.
Why is that unfortunate? Why is it "too bad for you?" Under Skinner's
scheme, behavior is simply an outcome of converging external forces. It
has no value; nobody wishes the situation to be any different from the
way it is. Nobody wishes anything.
And if the citizens find being whipped and chained punishing,
those behaviors of theirs which lead to the cessation of whippings
and chainings will be strongly reinforced.
.. and the behavior of the punishers that crushes the resistance by
intensifying the whipping and chaining will also be strongly reinforced.
On the other hand, perhaps the resistance will prove punishing to the
oppressors, so they will cease their oppression. Or the oppression will
punish the citizens so they no longer protest. Seems to me that no
matter what happens, you have it covered.
----------------------------------
Under PCT, you can't "deliberately" do anything either. You can
only behave as your reference signals and current perceptual inputs
command. Yes?
Under PCT, all behavior is deliberate: that is, it is produced in order
to accomplish the purpose we see being accomplished. If you ask where
the reference signal comes from, you find it comes from higher systems,
more removed from the environment. The whole system of reference signals
arises as part of a process built into the organism that maintains its
internal state in a match with a set of inherited criteria. The ultimate
cause of any behavior or any learned organization is therefore to be
found not in the current environment or in the history of a single
lifetime, but in the evolution of the species.
--------------------------------
Because Skinner emphasized planned environments as a means to
"control" behavior, many have ascribed to Skinner a right-wing,
almost fascist view. Dinsmoor informs us that the opposite was
true.
Actually the idea of planning a society for the good of its members is
also a left-wing view, viz the history of communism in the Soviet Union.
But political leanings are not the issue here: the issue is that under
Skinner's explanation of behavior, no organism ever plans anything. If
he ever spoke of planning an environment, he was violating his own
theoretical principles. Organisms do not plan: they emit behaviors at
random, and the environment selects some of those behaviors and
suppresses others. Of course Skinner may not have considered himself to
be an organism ...
Question: If you believe in Freedom and Dignity, in reducing
punative practices and increasing those that lead to a happy and
productive life, but your science has convinced you that behavior
is determined by the environmenal contingencies, what do you
recommend be done?
You pick a theory that allows an organism to have preferences for a
perceived state of affairs, and to behave according to the difference
between what is perceived and what is preferred to be perceived. Short
of that, there is nothing you can do but contradict yourself.
If behavior is determined by environmental contingencies, that means
your behavior, too. You have to give up the goals of freedom and
dignity, and of reducing punitive practices, and of doing things that
lead to a happy and productive life, because your science tells you that
goals are figments of the imagination, superstitions, throwbacks to the
days of religious oppression of science. In reality, your science tells
you, such goals are futile epiphenomena having no significance in the
determination of behavior. If you want to be completely consistent in
your theoretical views, you must dismiss all prescriptive statements and
simply wait for the environment to shape your behavior as it will.
On the other hand, if you aspire to freedom and dignity, and dislike
punitive measures, and generously wish for all human beings to have
happy and productive lives, then the only way to maintain your theory is
to become unconscious of saying and doing things that are supposed to be
irrelevant under your theory. You simply fail to notice that while all
other organisms behave because of schedules of reinforcement, you alone
have goals and aspire to reach them.
Skinner would NOT have replied, "So what?" because (in his view)
his genetic endowment, reinforcement history, and current
environment determine that he will attempt to so something to
improve society as he views it.
But Skinner does not "view" anything, nor does he "attempt to do
something", nor is there any purpose in him to make society into
something other than what it is. If Skinner lives by Skinner's theory,
Skinner is a nonentity, a passive bit of matter being moved around by
its surroundings. Skinner's humane leanings are in direct conflict with
his intellectual understanding of how behavior works. There is no way
out of it.
------------------------------------
Bruce Abbott (950710.1225 EST) (replying to Bill Leach) --
In its broadest sense, this says only the obvious, which is that
behavior depends not only on the current circumstances (feedback
functions, disturbances, etc.) but on past learning. This
assertion is no different than what PCT would hold.
Glad you brought this up. In fact there is an immense difference between
what PCT holds about learning and what EAB holds. Under PCT, BEHAVIORS
ARE NOT LEARNED.
Psychology has always studied behavior in terms of events. The rat in
the maze turns left or turns right. The animal on an FR-10 schedule
learns a high probability-density of emitting a keypress. A rat learns
to press the left key when the green light is on, and the right key when
the red light is on, and no key when no light is on. The student
presented with "2 + 2 = " learns to emit "4."
This is not how a control system works.
A control system is not a list of behaviors elicited by specific
stimuli. It is a system made of _physical processes that create
relationships among variables_. A description of a control system does
not make any reference to specific values of variables. It refers only
to functional relationships among variables that are independent of what
values the variables happen to have.
As a result, the only effect the past has on the present is carried in
the forms of functions that make up a control system. Once these
functions are defined for a specific control system, we know how the
system will behave under ALL circumstances, whether similar to those
that occurred in the past or novel.
There are many pasts that could result in a specific organization of a
control system. Once that organization exists, how it got that way is of
no further importance. Its behavior arises out of its _present_
organization, and the interactions of that organization with the
_present_ environment. PCT tries to explain behavior in terms of present
organization, not past history.
--------------------------------------------
So reinforcement theory is untestable because it appeals to
previous learning as a determinant of behavior but PCT escapes from
this conundrum because it appeals to the "influence" of past
experience through some other mechanims not stated? I don't
follow...
What I just said is the explanation. The goal of PCT research is to find
a model with an organization that will reproduce present behavior in the
present environment (including present reorganizing effects). This model
makes no reference to past events. We do not explain behavior in terms
of specific events in the past; we explain it strictly in terms of
present-time organization. As I said, there are many historical paths
that can lead to the same organization (through whatever adaptive or
reorganizing processes we postulate). But only the present organization
has any influence on behavior.
-----------------------------------------------------------------------
Martin Taylor (950710.1315) --
Unless I totally misread you (a distinct possibility), you are
asserting that subjective experience of one's own hierarchy--
perception OF perception rather than simple perception of the
environment--is our main access to the organization of the control
hierarchy. This may be all well and good for one's own experience,
but where does that leave you when someone else claims to have a
different experience that suggests a different kind of
organization?
I am not talking about perception of perception, but of perception,
period. The world you see around you IS (according to PCT) the world of
perceptual signals. What you are looking at is the set of all signals
emitted by all perceptual functions at all levels. You are not looking
at the causes of perceptual signals, but at the perceptual signals
themselves. That is how they look (,feel, taste,...).
Therefore, if you want to understand how perceptual signals are related,
look at how the elements of the experienced world are related.
-----------------------------------------------------------------------
Bruce Abbott (950710.1505 EST) (replying to Rick Marken) --
You are trying to evade the question by hiding under this
smokescreen of irrelevancies. If I recall the control system
correctly, error between reference and controlled perception brings
about action. The control system knows nothing of disturbances.
The action brought about depends on the structure of the control
system. If the system is properly constructed and tuned, the
action will tend to bring the controlled perception near its
reference value, i.e., the control system will control.
We have gone on too long leaving disturbances and reference signals --
but particularly disturbances -- out of the discussion. In fact, the
structure of a control system does NOT determine the action it will
produce. Actions are determined jointly by reference signals and
disturbances, acting through the existing structure. What is controlled
is the joint effect of actions and disturbances: the variable at the
_input_ to the control system, not the output.
When you allow arbitrary disturbances into the picture, you find that
there is no longer any predictable relationship between action and
consequence. The action may change; the consequence (to a good
approximation) will not. The behavior will change to keep the
consequence from changing.
If you want a crucial experiment, just set up an operant conditioning
experiment in which the reinforcer is subject to additive arbitrary
effects independently of the behavior, and arrange matters so the
organism has no way to sense the cause of these arbitrary effects
(except through the resulting changes in the reinforcer). As the
arbitrary effects change, you will find that behavior changes, too, by
just the amount needed to maintain the received reinforcement close to
its former level.
What remains constant is the consequence, not the behavior. A control
system is organized to control consequences, not behaviors. It will
produce whatever amount of behavior is needed to keep the consequence at
the level preferred by the organism. There seems to be a consistent
relationship between action and consequence ONLY when there are no
disturbances.
So you are wrong when you say
The control system knows nothing of disturbances. The action
brought about depends on the structure of the control system. If
the system is properly constructed and tuned, the action will tend
to bring the controlled perception near its reference value, i.e.,
the control system will control.
You are wrong because the action varies with disturbances; it is not set
by the structure of the control system. There is no particular action
that will bring the controlled perception near its reference level. Why
not? BECAUSE OF DISTURBANCES. You can't claim to understand how a
control system works until you see how such a system can _vary_ its
action quite automatically to counteract the effects of disturbances,
leaving the controlled variable essentially undisturbed. No fine tuning
is needed, no careful balancing of opposing forces. The gain of the
system, the ratio of action to error, can vary over a factor of 10 while
the action goes through its entire range and the controlled variable
changes by no more than a few per cent. Look at the right side of the
Motheral curve. Going from FR-1 to FR-2, the behavior rate doubles but
the reinforcement rate -- which is under control -- drops by only about
5%.
If this isn't clear, we need to do some simulation work to bring out the
basic properties of a control system, which are very different from
those of a reinforcement system.
------------------------------------
The argument Bill makes regarding the supposed inherent
contradiction in Skinner's system is no less inherent in PCT: _you_
can't deliberately do anything, because _you_ don't control the
reference levels and the disturbances.
But YOU ARE the perceptual signals, the reference signals, the error
signals, and the input, comparison, and output functions at all the
levels. When I say "you do" something, I mean that your brain is
responsible. This is the basic difference: is the environment the causal
agent, or is the brain?
If your only model of the brain is an input-output model, where sensory
inputs are simply routed by complex paths to the muscles, then the brain
becomes merely a passive relay of inputs to outputs. This is in fact the
kind of model that behaviorists have always believed in, and they have
been supported by biologists and neurologists. This is how the
environment is seen as still being in charge, even though its effects
pass through the brain.
But when we see the brain as a hierarchy of control systems, we see that
to a major extent it determines what its own inputs will be. The
environment does not have free influence over the organism's inputs,
because the organism's outputs are strongly affecting those parts of the
environment that affect the inputs. We find that reference signals in
the brain are the primary determinant of the inputs that are controlled.
And to explain the reference signals we have to go higher into the
brain, and ultimately must go to the inherited control systems that
maintain the basic structure of the organism itself. The ultimate
explanation of the reasons for behavior NEVER lies in the current
environment.
--------------------------------------
Rick:
You will also admit that there is, then, nothing one can do
about the situation. You would certainly not recommend that people
build Walden II because they cannot possibly do that unless the
environment happens, by chance, to select that behavior.
Bruce:
This is a non-sequitur. You would argue on this basis that a rat
can do nothing to obtain a food pellet because its behavior is
under environmental control. Yet reinforcement theory explains how
it can and very likely will acquire the ability to do so.
The rat itself has no influence over whether it will obtain food
pellets. All it can do is to emit various motor acts. It is the food
pellets that make some of those motor acts become more probable than
others, until finally the rat is pressing at a high rate and receiving
food pellets at a medium rate. Whether the rat likes to receive food
pellets is irrelevant; the reinforcing properties of the food pellets
make the rat press to obtain them whether it wants to or not. If bits of
gravel had the same reinforcing properties, they would make the rat
press the bar to obtain bits of gravel. All we know under reinforcement
theory is that food pellets happen to be reinforcing and bits of gravel
do not (for rats; bits of gravel can in fact reinforce some bird
behaviors). But there's no particular reason for that. It's just that
some items in the environment have the capacity to reinforce certain
behaviors in certain organisms, and some do not.
Bruce:
By simple extension Skinner can hope to alter the environment in
which people operate as a means to improving their lot in society.
No, to the contrary, he can't hope for anything. He will either behave
to alter the environment or he will not. He can find out which will
happen only after the fact. Skinner himself said that the only way he
deduces that he is going to mail a letter is to find himself walking
toward a mailbox with a letter in his hand. And to speak of doing one
thing "as a means" of doing something else is only to say that doing one
thing will (if it happens) have other consequences (if nothing
interferes). In no sense does one do things _in order to_ have other
effects. And there is certainly no concept of producing a consequence
_in spite of_ interfering effects.
(Whether he would opt to do that, however, would depend on his
inheritance, prior experience, and current circumstances.)
Again, wrong language. Skinner has no options. Inheritance, prior
experience, and current circumstances determine what will happen,
leaving no alternatives for Skinner to consider.
One way he might hope to alter the environment is by writing Walden
II, which, when read, might alter how people behave toward each
other.
Uh,Uh. Hoping can't affect anything: it's a mental state. Whether
Skinner will write Walden II is beyond his control; it is the
environment that determines this, that had already determined it by the
time Skinner started emitting typing movements. And Walden II was
certainly not written to bring about any new state of affairs. That
would be goal-directed behavior, which does not occur.
Furthermore, you are assuming that people are just sitting around
waiting for the environment to act on them. This is a gross
oversimplification of reinforcement theory (for one thing it
ignores the effects of prior experience) and it is this
simplification and not the theory which leads to the apparent
absurdity you point out.
It doesn't matter what people do; all that matters is what the
environment does, because ultimately everything people do is merely what
the environment has done to them. Prior experience merely reduces to
prior states of the environment (there is, of course, no such thing as
memory).
-----------------------------------------------
This is why I stated earlier that I thought the problem might be
definitional. Who is this "you" to which "you" refer? If "you" is
some mythical being inside your head who is "setting goals" etc.
then I would have to disagree. "You" can't control anything,
because you don't exist.
That's an interesting, and logical, conclusion. How does it feel not to
exist?
In fact, there most probably is a being in your brain setting goals and
so forth. This being is you. It is also your brain. There may be more to
you than your brain; I don't know and neither do you. But it is certain
that your brain can set goals, choose, think, and do all those forbidden
things. There isn't any little man in the brain, of course; the little
man is the WHOLE BRAIN. And since you are your brain, you can set goals
and so forth. The hierarchical model is a not-too-bad guess about how,
physically, this is accomplished at many levels.
The problem with this whole external-control approach is that it leads
ultimately to absurdities of the above type. Logically, we do not exist.
We can prove by careful reasoning that reason does not exist. We can
convince ourselves beyond doubt that there is no self to convince. Worm
Orouboros, munching away at its own tail and disappearing in a puff of
green smoke.
----------------------------------------
Skinner was aware of the dangers of "mentalism" to a science of
behavior and did everything possible to exclude what he took to be
mentalistic (including "cognitive") concepts.
What dangers are those? Skinner had a supersitious fear of mentalism,
but he had no rational arguments against it. What we have learned from
computers and other electronic systems tells us that there is no basic
difference between "mental" and "physical." We are quite used to seeing
artificial devices do thought-like things, and spontaneously generate
organized actions. As far as I can see, there is no longer any mystery
about mental phenomena; they are simply the operation of a signal-
handling system inside the brain. What's the problem? Skinner wouldn't
have known any of this, of course. He hadn't the education.
----------------------------------------
Actually, Rick, I'm not defending reinforcement theory, though I'm
sure it appears that way to you. But when I hear you or Bill or
anyone else making statements regarding reinforcement theory which
I perceive to be erroneous, I think the only intellectually honest
course of action is to disagree, even if that creates discomfort
and even suspicion as to my "real" views and motives. I guess I'm
just an iconoclast.
Understood. However, when you present the arguments from the side of
reinforcement theory, you also seem to adopt the sloppy habits of
thought that go with it -- the "have your cake and eat it too" attitude
that allows full freedom to speak in purposive terms, conveniently
forgetting the basic assumptions when sticking to them too literally
would result in an absurdity. The arguments you present have been around
for a long time; they've been worked over and agreed to and adopted
almost verbatim by many behaviorists. But people trying to prove a point
can be too easily persuaded by their own arguments, and can too easily
obtain acquiescence from colleagues who want to prove the same point.
What matters most is proving the point; what matters less is doing so by
impeccable logic. After all, if you know that the answer is right,
what's the point of going through a lot of labor to prove it's right?
--------------------------------------------
I know it's the wrong theory, but saying that it's wrong because it
can't predict the apparent position of Mars when it can isn't going
to do anyone any good.
The considerations relating to the Motheral curve will, I think,
eventually show that reinforcement theory can't predict the position of
Mars, except at just the right time of the synodic year. I believe that
to a considerable extent, the basic concept of reinforcement theory has
been a self-fulfilling prophecy: the experimental conditions have been
arranged (quite unconsciously) to make it seem to work. The exposure by
Motheral and others of unexpected relationships are like discovering
that you have been explaining the motion of Mars only when it was in the
retrograde part of the cycle, an illusion created by special viewing
conditions and the exact opposite of what is observed most of the time.
I know that you still find it unbelievable that this could have
happened, but I think that a search through the literature will show
that it did indeed happen. If it hadn't happened, nobody would be under
the impression that reinforcement "maintains" behavior.
That aside, there are still conditions we haven't explored: control in
the presence of unpredictable disturbances. Here I think we will find a
planet that moves in a way that is not only unpredictable by
reinforcement theory, but which collides with the basic thesis and
destroys it.
-----------------------------------------------------------------------
Bill
P.S. I agree this this whole exchange should be repeated on the net. The
copyright problem could be solved with a phone call and a follow-up
letter, or the listeners could be referred to the book. You guys decide
who will do it, OK?.
-----------------------------------------------------------------------
Bill