[Martin Taylor 930830 11:00]
(Rick Marken 930829.2030)
Why, for example, do people vary their
beliefs about the truth of propositions or the probability of the correctness
of implications. Studies of these kinds of things have been done -- but
usually in the "wrong" way, from a PCT perspective (eg. results based
on averages over subjects, individual results too noisy to be useful, etc).
There are two orthogonal problems here, which this language tends to mix up.
Many people who understand PCT have a particular position with respect to
the use of statistics in evaluating experiments. This view can be simplified
(caricatured) as "it's of no interest unless it works for each individual
100% of the time." All measures, whatever they may be, must be taken
within an individual and never across individuals.
It is true that most psychological studies are reported using statistics
in a way that is philosphically meaningless. They use significance tests,
for example, or use the trend of group means as a statistic representing
the mean of trends within the individuals of a group. These practices are
simply wrong, PCT-view or no. Significance tests tell you nothing more than
whether your experiment was sufficiently sensitive to detect a difference
that MUST always exist, and the failure to find "significant" results in each
of a sequence of experiments has often led to claims that the sought effect
"has been shown in repeated experiments not to exist," when in fact the
experiments considered together give strong evidence that the effect is
consistent and important.
These problems have nothing to do with PCT. They been discussed in the non-PCT
literature for at least as long as I have been aware (see, for example,
Edwards, Lindeman, and Savage, Psych Review around 1957--reference strictly
from memory). So this issue has nothing to do with PCT as such. It is my
view that the studies of PCT should benefit from a proper use of statistics
even more than do non-PCT studies. Why more? Because I believe PCT to
provide a more nearly correct view of psychological functioning than do
studies carried out in other conceptual worlds.
The second issue is that the studies have been done in the wrong way from
a PCT perspective, because they have not considered how the concepts
studied relate to the control of perception. One can think of "degree
of belief or disbelief" in a PCT light. A perceptual signal that would
create a significant error (deviance from its reference signal) "should"
generate no output if its reality is disbelieved, for example. It may
be the degree of belief that is the controlled variable.
ยทยทยท
==================
Within the Layered Protocol theory, there are three interlinked controlled
propositions relating to the sending of a virtual message, and it is the
degree of belief about these three propositions, and about the partner's
beliefs about them, and about the partner's beliefs about one's own beliefs
about them (three levels only) that are controlled (9 variables).
(The 3 propositions are, roughly:
P1. The recipient has made an interpretation of the virtual message;
P2. If P1, the interpretation is adequate;
P3. It is not worth continuing to try to improve the recipient's
interpretation.)
In a continuous control system, P3 is unnecessary, since it is a consequence
of the reduction of the error signal to zero, which can happen because P2
is strongly believed, but it can also happen because a higher-level virtual
message has its own P3 satisfied.
There is a complex system of algebra that we haven't fully worked out
relating to the interrelation of the 9 degrees of belief (Originator
strongly believes that recipient weakly disbelieves that ...). Any
assistance in improving it would be welcome. I can send an unpublished
book chapter to anyone interested in seeing if they can help.
Martin