Control of Perception: Fact or Theory?

[From Rick Marken (951110.0900)]

I said (951108.0800)] --

I have been arguing...that Powers' genius was the "discovery" of the FACT
than behavior IS control...Indeed, the title of Powers' book (Behavior: The
control of perception) is a statement of fact, not theory. Behavior IS the
control of perception. Control theory explains how this works.

Chris Cherpas (951109.0947 PT) --

I'm certainly seeing the same kind of in-group bravado in PCT statements of
the form "controlling perceptions is a FACT..."

Tom Bourbon (951109.1408 CST) replies --

When we are careful, we say that _control_ is a fact and control theory (a
theory of control) explains control

Bill Powers (951109.2040 MST) replies --

Well, you don't hear that from all of us. Control is a fact; control of
perception is a theory, and an experience.

I was being a bit incautious when I claimed that "control of perception" is a
fact. After some cautious reflection, however, I am prepared to argue that I
was right: control of perception is a fact.

When we say that control is a fact, we mean that we have discovered some
variable aspect of the external environment (such as the cursor in a
tracking task) that passes The Test; the variable is under control. But we
know from physics and physiology that its not the external variable (the
cursor) itself that is controlled; it is a neural representation of that
variable. In a tracking task, it's not the cursor itself that is controlled;
it is a neural representation of the optical effects of the cursor on the
retina that is controlled. I think it's fair to call this neural
representation a "perception". In this case, the term "perception" doesn't
carry the theoretical implications that it does in PCT (where "perception"
is defined as a neural current in an afferent neuron; this is a theoretical
statement about the _nature_ of the neural representation of the controlled
environmental variable) but I think it's a fair way to describe what we
know about the facts of the situation; that what is controlled is a neural
representation ("perception") of an external variable.

Even though we can't "see" the neural representation of the controlled
variable, I think we can consider its existance a fact in the same way that
we consider the existance of the (unseen) computer link between control
handle and cursor display to be a fact. The neural representation of the
controlled variable is a fact based on other facts -- the facts of optics and
retinal physiology; the computer link between handle and cursor is also a
fact based on other facts -- the facts of electronics and computer
engineering.

The idea that "control of perception" is a fact is also implicit in The Test
for Controlled Variables itself. Once a controlled variable (such as cursor
position in a tracking task) has been identified, we make sure that it is
controlled by the subject by interfering with the subject's ability to
perceive the variable (for example, we hide the cursor behind a cardboard
shield). If the subject is controlling the variable (cursor position) control
will be lost when the variable cannot be perceived. So The Test is based on
the assumption that the subject is controlling a perception (neural
representation) of an environmental variable.

The reasons for claiming that "control of perception" is a fact also apply to
cause-effect theories of behavior, which could be called "control by
perception" theories. Conventional psychologists (who accept the facts of
physics and physiology) know that a "stimulus" (independent variable) can
only have an effect on behavior via physical effects on the subject's
sensors, effects that are transformed into neural representations.
Reinforcement theorists, for example, who claim that food is a "reinforcer"
know that whatever reinforcing effects the food has, it must have those
effects via the organism's sensors; it is the effects of food on the retina,
the receptors in the mouth, the receptors that detect physiological effects
of the food, and so on that have an effect on the organism. Same is true of
conditioned and unconditioned stimuli; Pavlov knew that it must be a
perceptual representation of these stimuli that has an effect on the
organism, not the stimuli themselves; you can't side-step the sensory input
to the organism.

So, if the word "perception" is taken to mean "neural representation of
environmental variables" then I think everyone (conventional and PCT
psychologist alike) would agree that environmetal variables exist for
organisms only as perceptions (neural representations).

The difference between conventional and PCT psychologists concerns the
_factual_ role of perception (representations of external variables) in
behavior. Conventional psychologists would say that behavior is (in fact)
caused or selected by perception; PCT psychologists would say that behavior
is (in fact) the control of perception.

Best

Rick

[Martin Taylor 951110 13:30]

Rick Marken (951110.0900)

In a tracking task, it's not the cursor itself that is controlled;
it is a neural representation of the optical effects of the cursor on the
retina that is controlled. I think it's fair to call this neural
representation a "perception". In this case, the term "perception" doesn't
carry the theoretical implications that it does in PCT (where "perception"
is defined as a neural current in an afferent neuron; this is a theoretical
statement about the _nature_ of the neural representation of the controlled
environmental variable) but I think it's a fair way to describe what we
know about the facts of the situation; that what is controlled is a neural
representation ("perception") of an external variable.

I don't think you need go even so far as to commit yourself to a "neural"
representation. All you need is that there be some internal effect that
can be seen as having an associated magnitude value--any scalar quantity
will do, whether chemical, electrical, or mechanical. To go further, and
to assert that the relevant "perceptions" are neural, is both to restrict
the domain of PCT unnecessarily and to lay your "fact of control" open
to a falsification to which it should not be subject.

PCT seems to be valid also for organisms that don't have neurons. At least
I never thought that the controlled variable had to be a "neural current."
I thought that was just an exemplar, showing that the theory could have
direct correspondences in a nervous syetem.

Martin

<[Bill Leach 951110.22:55 U.S. Eastern Time Zone]

[Martin Taylor 951110 13:30]

Another nice posting Martin and a useful reminder.

-bill