Cause and effect, reinforcement theorists

[From Rick Marken (951125.1845)]

Bruce Abbott (951125.1510 EST) --

If an independent variable is being manipulated for experimental purposes,
it cannot be changing naturally, as then one would not have an experiment
but only a type of correlational design. A weakness of purely correlational
designs is their inability to prove that the relationships they disclose
are causal.

Since I get in trouble with Bill P. when I question your understanding
of PCT methodology, I'll take this opportunity to question your under-
standing of conventional methodology.

You are saying above that correlation does not imply causality because
the independent variable is left to vary naturally in the correlational
design but is actively manipulated by the experimenter in the experimental
design. This, of course, is incorrect.

What distinguishes experimental from correlational design is "control
of extraneous variables" -- where "extraneous variables" are all variables
except the independent and dependent variables in the experiment. Extraneous
variables are controlled (held constant) in experimental design; they are not
controlled in correlational design. When extraneous variables are controlled,
it can be "safely" concluded that variations in the dependent variable were
caused by variations in the independent variable and not by variations in any
extraneous variable (of course, we know that this "safe" conclusion is not
safe at all when the independent variable influences a variable controlled
by a control system).

The distinction between experimental and correlational designs has
nothing to do with who or what caused the variations in the independent
variable. It is perfectly possible to do an experiment in which
variations in the independent variable are caused "naturally" rather
than by the actions of the experimenter. It is also possible to do
a correlational study in which the independent variable is manipulated
by the experimenter. Correlation does not imply causality but this is
not because a human didn't manipulate an independent variable; it's
because a human didn't control all extraneous variables.

Bruce Abbott (951125.1820 EST) --

when you suggest that reinforcement theorists view reinforcement as
something residing in the reinforcer rather than in the relationship
between the perceptual effect of the reinforcer and the organism's
physiological state, you misstate their position, and they have every
right to take you to task for it,

Please post a reinforcement model that does not attribute to a reinforcer
an ability to strengthen that is completely independent of the state
of the organism.

Reinforcement theorists would be just as perplexed and irritated by your
attribution to them of "animism," which simply arises from a definitional
difference in the use of the word "control" and has no real substance.

If this is true, then why do reinforcement theorists (and their apologists)
so strongly resist our definition of control? If we're dealing with a
definitional difference of "no real substance" why do reinforcement
theorists react so strongly when we say "organisms control reinforcment".
Why don't they want to try to understand what we mean by that? If
it's just definitional, why are there no studies done by reinforcement
theorists that can even be construed as being about determining what
organisms control?

There are fundamental differences in approach and framework to discuss;

We've discussed them. Reinforcement theorists don't seem to like the
conclusions.

these silly attributions deflect attention from what control theory has to
offer and simply put what should be the intended audience into a defensive
stance, in my opinion.

Please state as clearly and simply as possible what you think control
theory has to offer reinforcement theorists?

Best

Rick

[From Bruce Abbott (9527.1650 EST)]

Rick Marken (951125.1845) --

And you were beginning to think I'd never reply to this one! (;->

Bruce Abbott (951125.1510 EST)

If an independent variable is being manipulated for experimental purposes,
it cannot be changing naturally, as then one would not have an experiment
but only a type of correlational design. A weakness of purely correlational
designs is their inability to prove that the relationships they disclose
are causal.

Since I get in trouble with Bill P. when I question your understanding
of PCT methodology, I'll take this opportunity to question your under-
standing of conventional methodology.

You are saying above that correlation does not imply causality because
the independent variable is left to vary naturally in the correlational
design but is actively manipulated by the experimenter in the experimental
design. This, of course, is incorrect.

No, I'm not saying that, and of course what you say I say is incorrect. The
whole point of an experiment is _control_. If you allow the independent
variable to vary on its own, you have no idea what other variables are
varying with it. For example, I may find that certain organisms are more
active in the sunlight than out of it. Is it the level of illumination that
is responssible, or the increase in temperature that accompanies the
sunbeam? If you manipulate the independent variable yourself, you are able
to control both it and any potential correlates you believe need to be
eliminated so as to prevent confounding. You can pass the sunlight through
an IR filter, for example, and then vary the intensity of the beam. It's
not that _I_ vary the intensity of the beam that's important; it's that in
so doing, I am able to vary it is such a way that _only_ the intensity of
the beam varies.

Bruce Abbott (951125.1820 EST)

when you suggest that reinforcement theorists view reinforcement as
something residing in the reinforcer rather than in the relationship
between the perceptual effect of the reinforcer and the organism's
physiological state, you misstate their position, and they have every
right to take you to task for it,

Please post a reinforcement model that does not attribute to a reinforcer
an ability to strengthen that is completely independent of the state
of the organism.

Would it be O.K. if I just described one? How about the classic (and
defunct) Hullian model, in which the change in habit strength following
reinforcement is a function of drive, which is taken to be an internal state
of the organism? The model is multiplicative; when drive is zero, there is
no strengthening.

Reinforcement theorists would be just as perplexed and irritated by your
attribution to them of "animism," which simply arises from a definitional
difference in the use of the word "control" and has no real substance.

If this is true, then why do reinforcement theorists (and their apologists)
so strongly resist our definition of control? If we're dealing with a
definitional difference of "no real substance" why do reinforcement
theorists react so strongly when we say "organisms control reinforcment".
Why don't they want to try to understand what we mean by that? If
it's just definitional, why are there no studies done by reinforcement
theorists that can even be construed as being about determining what
organisms control?

You seem to be misconstruing my meaning. When someone in operant
conditioning says that the environment controls behavior, you put words into
his or her mouth by insisting on your own definition of control, which
includes the concept of goal or intention. But the EABer only means by this
that behavior changes as a function of environmental inputs. The silly
argument that ensues from your claim resides in the different definitions
for the word "control" that you and the EABer apply. Why, if inanimate
things in the environment control an organism's behavior (using the PCT
definition of control), then they must have goals, and be striving to bring
them about. Hence the claim that EABers are animists. But this is silly.
The EABer's definition of control includes no such concept, and so the EABer
is not acting as an animist as he or she uses the term. It all comes down
do the difference in definition.

I certainly do NOT mean that PCT and reinforcement explanations for the
purposiveness of behavior come down to a mere difference in how the two
camps define "control," which is what I'm guessing you think I meant. The
central difference between these two views is this: In PCT, control systems
control, and behavior is just a byproduct, a means to that end. Organisms
do have real intentions, goals, purposes, represented as internal signals.
In reinforcement theory, intentions, goals, and purposes are illusions, for
the same reason that the apparent purposiveness of evolution is an illusion.
In this view, the organism is constructed (because of its evolutionary
history) in such a way that certain kinds of events, under certain
conditions, serve as reinforcers. Behavior that produces such events
becomes more likely, because bringing about such events in the organism's
evolutionary history has contributed to greater reproductive success. Under
the selective effects of reinforcement, those behaviors that produce those
reinforcers tend to predominate over those that don't. The result is that
the organism _appears_ to be doing the right thing (as required for its
survival and reproductive fitness), as if achieving that outcome were its
goal, just as the turtle wears a hard shell, as if it did so _in order to_
gain some protection thereby.

Now THOSE are fundamentally different explanations for the purposiveness of
behavior, and not merely a definitional quibble!

Now, back to your questions:

If this is true, then why do reinforcement theorists (and their apologists)
so strongly resist our definition of control?

Probably because they have already defined the term a different way, and
don't want to change their usage of the term just to suit you.

If we're dealing with a
definitional difference of "no real substance" why do reinforcement
theorists react so strongly when we say "organisms control reinforcment".

I don't think they do. It is an aphorism in EAB that control (EAB
definition) is reciprocal: reinforcement affects behavior; behavior affects
reinforcement. What they're probably reacting to is your calling them
animists. (;->

Why don't they want to try to understand what we mean by that?

Good question. Some have, but have misunderstood. Others may be so
convinced that the traditional approach is valid that their minds are not
open to alternatives. Certainly at this stage of the game, reinforcement
theory appears to explain a wide range of phenomena of conditioning better
than PCT does. Maybe that's why some are not ready to be convinced. Others
have never heard of PCT, despite your efforts.

If
it's just definitional, why are there no studies done by reinforcement
theorists that can even be construed as being about determining what
organisms control?

Here you are speaking of something more substantitive than the definitional
silliness I was complaining about. If (as EABers believe) the appearance of
control emerges from differential reinforcement, then there is nothing to
study except differential reinforcement. Or to put it another way, every
study of differential reinforcement is a study of control, so understood.

There are fundamental differences in approach and framework to discuss;

We've discussed them. Reinforcement theorists don't seem to like the
conclusions.

One problem is that if you convince a reinforcement theorist that PCT is
correct, he or she is no longer a reinforcement theorist. I'd like to note
that _I_ like the conclusions, but then I don't count because I'm no longer
a reinforcement theorist! (More accurately, I was never entirely
comfortable with the whole gig in the first place, but know all the
mantras.) As for the rest of them, too bad. They're all nearing retirement
anyway.

these silly attributions deflect attention from what control theory has to
offer and simply put what should be the intended audience into a defensive
stance, in my opinion.

Please state as clearly and simply as possible what you think control
theory has to offer reinforcement theorists?

A whole new way to view behavior, reinforcement, and purpose.

Regards,

Bruce