Rick refers to q.i as a hypothesis and as a perception of the researcher (or the researcher’s surrogate).Â
“q.i is a perception in the researcher (or the researcher’s surrogate, such as a computer), which, when there is sufficient evidence from the test, is presumed to correspond to the perception, p, that S is controlling.” -Rick
The way his sentence is written makes it seem as though q.i may, at the same time, not correspond to the perception, p, that S is controlling, as well as still be a perception in the researcher. What is a false hypothesis in PCT? Is it a perception? an imagination?
···
On Sat, May 26, 2018 at 12:03 PM, Richard Marken csgnet@lists.illinois.edu wrote:
[Rick Marken 2018-05-26_12:03:32]
Martin Taylor (2018.05.25. 23 13)–
MT: The question that underlies this often-repeated statement is whether
a particular function of environmental variables corresponds to a
structure that exists in Real Reality (RR).
RM: No, I think this is not really relevant to my point I am making about the relationship between q.i and p. I’ll try another approach.
RM: I think the difference between p and q.i becomes obvious when you actually do research on PCT. The “coin game” described in B:CP is one example of PCT research that illustrates the relationship between q.i and p rather nicely. E’s hypotheses about the aspect of the coins that S is controlling are E’s own perceptions of different aspect of the coins – different q.i’s. For example, one hypothesis about the aspect of the coins being controlled is the pattern of the coins (straight line, Z pattern, etc); another is the relative size of the coins; another is the relative value of the dates on the coin. All these hypotheses about q.i are E’s perceptions that might correspond to the perception, p, that S is controlling. The goal of the test is for E to find his own perception of an aspect of the coins, his own q.i, that corresponds to the aspect of the coins, p, that S is controlling. This happens when E finds that S corrects for all disturbances to q.i, the current hypothesis about the controlled variable.Â
RM: E’s perception of q.i need not be a “direct” perception of the hypothetical controlled variable, as it is in the coin game, where E uses his own perceptual system to see whether S is controlling the pattern, relative size, relative value of the dates, or something else about the coins. Indeed, in formal PCT research the hypothesis, q.i, about the perception, p, S is controlling is typically computed by a computer rather than E’s own perceptual systems. An example of this kind of research is our paper on intercepting toy helicopters:
https://www.dropbox.com/s/eymkj4bxuorpyuy/Chasin%27Choppers.pdf?dl=0
RM: This paper was meant to be an example of how to study behavior from a PCT perspective. The behavior under study was object interception – running to intercept a relatively randomly moving toy helicopter. Several different hypotheses about the perceptions that S is controlling in this situation were tested. These hypotheses are E’s perception, q.i, of the perception being controlled by S, p, just as in the coin game. But these hypotheses about p – these q.i – were defined by mathematical functions and computed by a computer. The test was done by putting different computed values of q.i into a model (as the perceptual variable, p) of S’s behavior to see which resulted in the closest fit to S’s behavior.Â
RM: There is another example of doing the test by having the computer compute the hypothetical controlled variable in Marken, R. S. (2014) Testing for
Controlled Variables: A Model-Based Approach to Determining the Perceptual
Basis of Behavior, Attention, Perception
and Psychophysics, 76, 255-263, which is reprinted as Chapter 4 in “Doing Research on Purpose” (https://www.amazon.com/Doing-Research-Purpose-Experimental-Psychology/dp/0944337554/). In that paper two different hypotheses about the perception controlled in a tracking task are tested. The two different hypotheses are two different mathematical functions that define for E the perceptions – q.i – that might correspond to the perception - p – that is controlled by S. Again, these hypotheses are tested using modeling.Â
RM: Another way to get a sense of the relationship between q.i and p is by doing the hierarchical control demo at http://www.mindreadings.com/ControlDemo/Hierarchy.html. In this case, you are S and the computer is E. The computer has three hypotheses – three different q.i – about the perception, p, you are controlling: one q.i is shape, another is direction of movement and the third is the sequence of shapes. The computer tests the hypothesis about the perception you are controlling by seeing which perception is best protected from disturbance, which is measured as the proportion of a trial that each hypothetical controlled perception – each q.i – is kept in a reference state.Â
RM: I hope you can see that when you actually get down to rolling up your sleeves and doing PCT research the relationship between q.i and p becomes quite apparent: q.i is a perception in the researcher (or the researcher’s surrogate, such as a computer), which, when there is sufficient evidence from the test, is presumed to correspond to the perception, p, that S is controlling.
BestÂ
Rick
That is something we can
never know for sure, but what we can know is that if we have a
perceptual function that does not correspond reasonably closely to
some true property of RR, controlling it isn’t going to do much to
help our intrinsic variables maintain us in good condition, so
reorganization will tend to remove it more quickly than it would
remove perceptual functions that do correspond to real structures in
RR. Controlling those variables has a better chance of producing
side-effects that are valuable for keeping intrinsic variables in
good condition. The same goes in spades for perceptual functions
that might have been developed by evolution because they are likely
to correspond to structures that have persisted in RR over
evolutionary time.
The takeaway is that transient perceptual functions (and arbitrary
ones programmed into robots) have no necessary relationship with
structures in RR, but persistent perceptual functions probably do.
Going back to what I think lies behind Rick’s statement, all
perceptual functions produce perceptions by manipulating a bunch of
variables. Some of them have outputs that change in the same way
that a property of the environment does when a particular input
variable is changed. For these, q.i is likely to be a value of a
variable in the environment, not simply a mirror of the perceptual
function output “p”. For the ones that don’t, there is no structure
in the environment that is constrained to vary as the perceptual
function output does when one of the input variables changes. For
these, q.i is a synthetic variable imposed on the environment by the
perceptual function, not a variable constrained by a structure in
the environment.
To make this concrete, suppose that in RR there is something that
produces the visual appearance we call a simple table, with four
legs and a top. Maybe there is a table there, but maybe the visual
appearance is a contrived illusion. You try placing the perceived
table top in a desired location. If the visual appearance of “table”
is the result of there being a real table “out there”, then when you
move the table top, all the legs will move as well, and they will
move in ways that cause the 3-D perceptual appearance of “table” to
be invariant. If the table is rotated, then the legs will move to
keep each one under the same bit of the table top, and so forth. On
the other hand, if it was an illusion, you might move the table top
and find that one or more of the legs does not come along. That
(non-)table was a construction created by your perceptual function.
The one that kept its 3-D shape might also have been, but it is much
more likely that the table was real than is the case of the one that
did not keep its shape.
The important concept here is that there are two feedback processes
involved. One is the immediate control of perception now, the other
is a slow reorganization loop that changes perceptual functions
according to what is really out there, bringing perceptual functions
more into alignment with Real Reality. In this loop Boss Reality
acts on the internal structure of the organism, tuning the
perceptual functions that produce the perceptions that are
controlled by influencing Real Reality. It was in this sense that
Powers called the influence of Real Reality on organisms “Boss
Reality”. Boss Reality determines what actually happens, and
perceptual control develops over the eons and over individual lives
toward a state in which perceptual reality approaches, but never
exactly matches, some part of Real Reality.
Martin
–
Richard S. MarkenÂ
"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery
RM: The problem is thinking of p as a function of q.i.
You have to remember that q.i is not an
environmental variable; it is a function of
environmental variables,
