The attractions of IV-DV

[From Rick Marken (930426.2100)]

Joel Judd --

Perhaps there's enough rejected papers to suggest an anthology to
some publisher under the title THE NOSES IN FRONT OF THEIR FACES:
IGNORANCE OF PURPOSEFUL BEHAVIOR IN THE SOCIAL SCIENCES or something
like it.

Beautiful title: I love the idea.

Joel Judd (930426.1130) --

So can we consider the setting an IV? Maybe in some general sense, but
the temptation in doing so is then to sit down and figure what
particular aspects of the environment I (researcher) can reproduce
or manipulate to produce the changes I predict.

You very interesting post put me in a philosophical mood regarding
IV-DV. Why is the IV-DV orientation to research is so seductive
and universally accepted in the behavioral sciences? One reason
it is, I think, because the systems DOING this research
(psychologists, sociologists, communications majors, etc) are
control systems themselves. It is natural for these systems
to want to take actions that produce reliable, perceptual
results -- that is, it natural for them to want to control;
many scientists want to control in this way. A physicist,
for example, might like to see if s/he can produce reliable
results (in terms of a voltage meter reading) by putting different
kinds of materials in an electrical circuit -- the scientist can
learn how to control the voltage. The materials used are the
IV and the meter readings (voltages) are the DV. A psychologist
might want to see if s/he can produce reliable results (in terms
of bar pressing rates) by using different substances as reinforcers;
the psychologist can learn how to control the bar press rate.
The substances used are the IV and the pressing rates are the DV.
Ultimately, the physicist and psychologist are trying to produce
controllable perceptions. The physicist is much more successful at
this than the psychologist usually. The reason for the comparative
failure of the psychologist relative to the physicist, I believe, is,
at least partly, because the physicist and the psychologist are BOTH
ignoring something that is only appropriately ignored by the physicist
-- the point of view of the system being studied.

The success of physics is a testament to the fact that it is
appropriate to ignore the point of view of the systems being
studied. According to PCT, the failure of the behavioral sciences
is precisely a result of the fact that the point of view of the
system is being ignored. The IV-DV approach is perfectly appropriate
in the study of physical, chemical and other non-living systems. It
is not appropriate for the study of living systems precisely
because it is not designed to learn the point of view of these
systems.

The goal of the behavioral sciences (according to PCT) must be quite
different than the goal of the physical sciences. The goal of
the behavioral sciences must be to learn about the system FROM THE
SYSTEM'S POINT OF VIEW. That's what the test for the controlled
variable is all about -- trying to figure out what the system is
trying to perceive. But it is hard for control systems (the one's
doing the research) to buy into this very new orientation towards
the study of systems. I think this is because 1) being control
systems themselves, these researchers naturally like to control; so
they like to be able to do things to a system (vary it's environment
or look at variations across several systems) and see if these
variations (IVs) produce predictable results. This is why, in the
rubber band demo, people typically get VERY interested in the
fact that you can make people move their finger in predictable
ways; the the fact that the person is doing this in order to keep their
own perception of the knot on target is of almost no interest at all --
what you typically get when you point out this obvious fact (so obvious
that it it has been completely ignored in the behavioral sciences) is
something like "well, of COURSE they are keeping the knot on the target,
dodo -- that's what you asked them to do". People love to control
because they are control systems. They are not very interested
(obviously -- given the reception of PCT) in the fact that other
people are controlling. Control systems care about what they can
"do" with other systems -- they don't seem to care much about
what the other systems are trying to do for themselves.

I also think researchers are not particularly interested in the
PCT (non - IV-DV) approach to research because 2) the IV-DV approach
works well enough. Control systems DO respond to disturbances so
you CAN act on them to produce predictable results. I think Greg
Williams once made an excellent suggestion about how to deal with
this -- just show that PCT lets you predict behavior better than
the IV-DV approach. That is, once you have identified a controlled
variable you can control a person's actions to a tee -- while
the IV-DV person is still computing his/her statistics.

PCT encourages a whole new perspective on what you want to
find out about living systems; it encourages you to abandon
the seductive fun of "making things happen" with people (by
manipulating an IV and watching a DV) in favor of trying
to learn what people are trying to make happen for themselves.
PCT gives a whole new perspective on what the "problems" are
in the behavioral sciences; the problems involve learning how
the system controls -- not how to control the system.

Best

Rick

Dear Rick,

I was reading this nice post about 2:30 am and could not but notice an
irony that it might be worthwhile pursuing in your next paper - if a
person is really interested in controlling (in the ordinary sense)
another then knowing what the other is controlling for is much more
useful, precise, and accurate as compared with sticking with
the idea that there is something in the environment that stimulates
action. As you accurately note those who do the RB exercise are more
interested in the "influence" aspect of the exercise than the "control"
aspect {in fact, one student in my class accused me of teaching how to
manipulate another with the RB exercise !} Since, everyone is interested
in control (in the ordinary sense) then maybe they would publish your
paper as a better way to do it. [Your problem will be hiding this from
Bill]

Regards, Chuck

[From Marcos Rodrigues (930427.1155 BST)]

Rick Marken (930426.2100) --

I'm sorry for your rejected paper. On the subject of IV-DV research,
I think you hit the target with your comments:

Why is the IV-DV orientation to research is so seductive
and universally accepted in the behavioral sciences?

...

I think this is because 1) being control
systems themselves, these researchers naturally like to control; so
they like to be able to do things to a system (vary it's environment
or look at variations across several systems) and see if these
variations (IVs) produce predictable results.

...

2) the IV-DV approach
works well enough. Control systems DO respond to disturbances so
you CAN act on them to produce predictable results.

Last month I went to Birmingham to visit my brother-in-law
who works in the neural sciences department at the Univ. of
Birmingham. His actual field of research is long and short-term
memory. The way they study memory is quite peculiar (to me,
anyway): they do it by slicing rat's brains and applying very
expensive drugs and electrical impulses on one side of the
slice (very thin, measured in microns) and observe what happens at
the other end. They measure lots of things, like how long
the potentiation persists, the effects of drugs and mixed drugs,
etc, but their experiments and their results are always described
according to a stimulus-response model. This is extended to the
whole of neural sciences, as some examples of their language will
confirm that:

David van Essen, Washington Univ, MO, on Neural Mechanisms
Mediating Pattern Vision in Primate Visual Cortex: "In addition,
responses to an optimal stimulus within the receptive field
can usually be modulated by textural patterns lying outside
the classical receptive field.... we presented cells with
non-Cartesian gratings, a novel class of stimuli that includes
hyperbolic patterns and polar patterns."

The same author on What's in a Receptive Field? Filters vs.
Feature Detectors Revisited: " Another line of thinking
emphasizes that neurons at all levels of the visual hierarchy
are typically tuned along multiple stimulus dimensions. From
this latter viewpoint, the primary objective should be to
identify the full set of stimulus dimensions that yield
well-defined tuning curves for each cell."

Stephen Redman, Australian National Univ, Canberra, on Long-Term
Potentiation at Excitatory Synapses on CA1 Pyramidal Cells:
"This analysis has been applied to EPSCs evoked in CA1 pyramidal
cells by stimulating small numbers of axons in stratum radiatum,
measured before and after conditioning stimulation which induced
LTP."

M.A. Goodale, Univ. of Western Ontario, Ontario, on Not One,
Not Two, But Many Visual Systems: "Moreover, many of the visual
control systems for the different motor outputs evolved as
independent, but interactive, input-output models."

However the importance of S-R models for neural sciences, the
hot stuff at the moment seems to be Central Pattern Generators
(CPG). Did you know that scratching is produced by a CPG?
And that CPG is an input to a control system? Shame on you.
Read this: R.E. Burke, National Institutes of Health, Bethesda,
MD, on The Organisation of Last-Order Interneurons Related to
Locomotion: "The three basic input sources that control spinal
cord motor output are: 1)sensory afferents (associated with
`reflexes'); 2) supraspinal descending systems (associated with
`voluntary' movements); and 3) central pattern generators (CPGs;
associated with repetitive, patterned movements like locomotion
or scratching)."

Can you find anything more outrageous than that? I'm glad
they are not spending MY tax money.

A.J. Simmers et al., Univ. of Bordeaux, France: "Motor
programmes driving rhythmic movements derive from neural
circuits (Central Pattern Generators) which are commonly
believed to consist of predefined assemblages of interconnected
neurones that are each dedicated to a different behaviour."

Paul S.G. Stein, Washington Univ, MO, on Motor Program Selection
in the Turtle Spinal Cord: "The turtle spinal cord has intrinsic
neural circuitry sufficient to select an appropriate motor
program in response to a specific cutaneous stimulus. Stimulation
of a site on the body surface can activate either a rostral, a
pocket, or a caudal scratch; stimulation of the dorsum of the foot
can activate limb withdrawal... These aspects were revealed
using electrical stimuli delivered to the cutaneous nerve that
innervates the ventral-posterior portion of the pocket scratch
receptive field. 3-Hz nerve stimulation elicited a robust pocket
scratch motor pattern."

Isn't revealing that the stimulation of the dorsum of the foot
can activate limb withdrawal?

Thinking about behavioural sciences, if people want to be
"scientific" about their subject, they must do what neural sciences
do. The truth is that you cannot negate the results of neural
sciences experiments, that is, by applying an electric current on
one side of a membrane, you will get some potentiation at the other
end. If you don't get it, you then immerse your test tissue in
some very expensive drug, do some meaningful measurements, like
index of facilitation of one drug compared to another, apply
stronger and longer stimuli, etc. The conclusion of such an
enlightening experiment is that the neural spike was a response
to your stimulation. I'm afraid that the impregnated S-R model
in the life sciences will last for quite a while.

Regards,

Marcos.