[From Rick Marken (950102.1645)]
Bruce Abbott (941231.1530 EST) --
It would be a very dull thing if all scientific research on the behavior
of living organisms were reduced simply to performing "the test" ad
nauseam.
Hmmm. I wonder why you don't feel the same way about the IV-DV
methodology that has been used, ad nauseam (and inappropriately) to
study living control systems for over a century. Inquiring minds
still want to know;-)
For example, consider all the work done by ethologists and
comparative psychologists over the past 60 years ...
Does the fact that it was _done_ prove that something useful was
discovered? Some of the observations made by ethologists and
comparative psychologists may be suggestive, but most the studies
with which I am familair don't tell us much about what we need to know --
what organisms control.
The functions of many of these behaviors in the life of the animal,
their development and dependence on experience, their necessary
physiological conditions and sensory "triggers"-- all were observed
and explored using many of the traditional methods of scientific
inquiry
Methods that have revealed the dependence of behaviors on sensory
"triggers" have kept behavioral science in the dark ("blind") for the last
100 years. It's time to take off the blindfold and see that organisims are
controlling their sensory input, not vice versa.
detailed in my text (which, by the way, includes much more
than just "IV-DV" methods).
You mean correlational? Quasi-experimental? Etc? Do any of these methods
recognize that organisms are controlling their own sensory inputs?
Can any of these methods reveal the existence of controlled perceptual
variables?
The methods of PCT can elucidate the underlying mechanisms of
control through which, for example, the male and female stickleback
conduct their elaborate ziz-zag courtship dance, but they could not
have predicted the dance in the first place.
No method can "predict" observations before they are made. More
importantly, PCT shows that observations of the _behavior_ of
a control system are, basically, irrelevant. PCT shows (before
any theory gets into the picture) that you can't tell what a
control system is doing by just looking at what it is doing. You can
only tell what a control system is doing by determining what it is
controlling. This fact is demonstracted by my "mind reading" program,
which is available on Dag's Demo disk. Matter of fact, my "Mind Readings"
book might be a useful addition to your PCT collection.
The courtship dance of the stickleback is something that these fish do;
but PCT says that what we see is a side effect of their controlling. The
fish cannot be controlling for a "zig-zag" pattern because it cannot get
outside of itself, which is where it would have to be in order to be able
to perceive (and control) this pattern. To find out what the fish is
doing, you have to determine what it is trying to control. That means,
I'm afraid, doing the test for the controlled variable(s).
PCT can't be used to explain of the behavior of the stickleback until we
know what the fish is doing (controlling). The observation of the zig-
zab dance is a suggestive observation. But it is just a start. The next step
is to determine what variables the fish might be controlling. Once we
know what variables are controlled, we can build a PCT model of the
stickleback and see if the model behaves like the fish.
It turned out that a-MSH suppressed these DVs in a dose-dependent
manner.
Again, there is nothing wrong with this observation per se except that
it tells us zilch about what the chicks might be controlling (or trying to
control) by chirping.
Again, a clear understanding of the underlying ECUs would help to
explain how a-MSH produces this effect,
This is putting the cart before the horse; you have not discovered any
controlling so why use a model of control to explain this phenomenon?
The observation alone suggests that no controlling is involved at all.
a-MSH dosage causes chirp supression. Where's the controlling?
but the discovery of the a-MSH effect itself did not require the
application of PCT methods.
If the a-MSH effect is real -- that is, if a-MSH dosage accounts for
something like 95% of the variance in chirping -- than you certainly
have discovered something. Most of the "effects" discovered by
conventional research methods account for considerably less than 99%
of the variance in the DV (I'd say a researcher would feel elated with
60%; in PCT we don't get excited unless were accounting for a least 95%
or so) so not only do they not reveal any controlling; they don't reveal
anything at all.
But even if the a-MSH effect is real, it is almost completely useless
(from a PCT perspective) since we have no idea how it related to what
the bird is controlling. For example, let's assume that the bird is
controlling the perceived intensity of its own chirping (this is
something to be determined by experiment, of course). Then the a-MSH
may reduce chirping because 1) it magnifies the birds perception of
chirping or 2) it changes the bird's reference setting for its perception of
chirping or 3) it zeros the gain of the chirp control system by paralyzing
the bird's chirp production system, etc. All of these hypotheses are
just wild and crazy kindda guesses until we know that the bird is,
indeed, controlling the perceived intensity of its own chirping.
I could go on to list hundreds of similar examples from other fields of
psychology, but I think I've made my point.
As you can see, I am not convinced. The examples you cite reveal nothing
about the controlling that the observed behavior is presumably (according
to PCT) a side-effect of.
Conventional research methods give the illusion (literally) of showing
something about the nature of behavior. It is a compelling illusion; if it
weren't, smart people would have not have kept deluding themselves
with it for over 100 years.
Adopting PCT does not automatically render these methods invalid,
One should adopt PCT only AFTER one understands the nature of
control. What renders conventional research methods invalid is the
FACT that organisms control. The methods described in your (and my)
research methods book are valid for the study of the behavior of living
systems only if those systems are NOT control systems. I can provide
evidence that they are; do you have any evidence that they are not?
In proclaiming that they [conventional research methods] are totally
inappropriate to the study of living control systems, you go much too
far.
Yep. That's what nearly everyone thinks (except the few of us who
understand PCT). It seems so _extreme_; how could we be so impudent as
to reject the entire edifice of behavioral research methodology?
Well, we don't do it because we want to; and we don't do it because we
"believe in" PCT. We do it because we know (for a fact) that organisms
exist in a negative feedback relationship with respect to their own
sensory input: organisms CONTROL. That's the fact, Jack. It's a fact that
we have demonstrated over and over again. It's a fact that has been
ignored, dismissed as unimportant or assumed to have been taken into
account in conventional behavioral science.
In order to be intellectually honest, once you accept the FACT that
organisms control you also have to also accept the fact that
conventional research methods are totally inappropriate to the study of
living systems. Here's why:
The basic assumtion of conventional behavioral research is that
o = f(s)
Output (the dependent variable, o) is a function of input (the
independent variable, s). This is why the journals are filled with
studies where independent variables are manipulated and their effect
on dependent variables is determined. This approach should tell you
something about f(), the organism function. If organisms are Z-systems
then f() tells you something about the nature of the organism.
If organisms are closed-loop control systems (which, of course, they are)
then:
o = 1/g(-s)
Dependent variables are a function of independent variables but the
observed relationship between these variables is the inverse of the
feedback function, g(), relating output, o, to a controlled variable, q.
The relationship between dependent and independent variable will be
noisy (like those found in real behavioral research) if o and s (a
disturbance) are among many variables related to a controlled variable.
But even if the relationship between independent and dependent variable
is nearly PERFECT (as it often is in operant research), that relationship
tells us nothing about the organism; it only tells us about a property of
the environment -- one that we could have observed directly, without
using the organism as an analog computer.
It's important to defer acceptance of PCT (theory) before one has a clear
understanding of the fact that it is designed to explain -- the fact of
control. Once you understand control, you will see that giving up on
conventional research methods is no more extreme than giving up on alchemy.
Of course, it's got to be a little rough emotionally if you've already
got a best selling alchemy text on the market;-)
Best
Rick