I think the point is that you can’t tell what someone is doing by watching what they’re doing. ;->Â But to be less flip about it, w
hen I am controlling a perception of program A, System B that perceives the program and perhaps controls its perception of the program (at least to the extent of “is it running or not”) is distinct from program A, the program that I am perceiving, andÂ
the real problem is that System B is controlling perceptions of the output of program A. Is System B executing the relevant if-then expression from program A (“square – if square then red – is the next shape red?”)? Only long enough to figure out the practicable cue-and-response sequences. During the learning phase System C uses programmatic control to analyse the consequences of the program structure (which was a given, in the instructions) and extract those criterial cues: if square, then red; if circle, then blue.
When I perform your demo, what I am controlling is a predictable sequence-pattern of shapes and colors. There’s more than one way to recognize the pattern so that I can press the spacebar when the pattern fails to occur. There’s more than one way to recognize it because there’s more than one way to generate it, or to generate a pattern that mimics it closely enough to serve the purpose. Maybe there’s more than one if-then program that would produce that observed pattern. The one that I offered might qualify. As a limiting case there’s a simple computer program which mimics the behavior but doesn’t involve if then " contingencies: a list of colored shapes which for a sufficient duration mimics output of the program (admitting no circle followed by red or square followed by blue), a second list that doesn’t comply, a random switching from one list to the other, and reversion to the ‘correct’ list with a bar press.Â
So as I said, what I think I am doing is recognizing and being alert for the two ‘fail’ sequences, circle followed by red color or square followed by blue color. That can be expressed by the program expressionÂ
To get there, I did indeed start by controlling a perception of the program that the computer is running to generate the changing shapes and colors, which was enabled by controlling perceptions of your description of that program. But I very quickly transformed that into a description of an algorithm for what I needed to do. While I am executing that algorithm, I am no longer controlling a perception of the program that the computer is running. Nor am I running if-then expressions from that program. I’ve reduced it to a level where I can control the inputs more efficiently–the sequence level.
Back to the core problem: you want to demonstrate that I am perceiving a program and controlling whether it is operating or not, but maybe you’re only demonstrating that I am perceiving criterial aspects of its output that indicate when it has stopped operating. I control those criterial perceptions so that I can intervene and start it again. Sort of like watching an assembly line, alert for known anomalies. The guy on the line might be the manager who knows how the line operates, or he might be Charlie Chaplin’s character just looking for the twisted widget. You can’t tell what he’s doing just by watching what he’s doing.Â
You want to demonstrate that we control program perceptions. Clearly we control with program perceptions. (Generally, we do this to figure something out.) And clearly we control program perceptions from a higher level or from within another program, in the sense of selecting the appropriate program to use for present purposes. But you have set this up as a problem of recognizing when another entity is or is not running a specified program. You want us to be the manager watching the doofus on the assembly line. Prior to the demo we are given a description of the program. From this we can figure out how we can tell when it is not running. The criteria by which we perceive that it has stopped running are not themselves perceptions of a program, they are perceptible features of the output of the program. To figure out those criteria we had to apply programmatic control to a perception of the program running in the computer. For that process, your demo does demonstrate control of the perception of the program as described, or so I think. But during the running of the demo a high score does not indicate that I was currently controlling that perception. It does indicate that my analysis was correct, i.e. arrived at the necessary criteria to perceive and control. But I could have arrived at it from the description of a different computer program with similar output.
···
On Fri, Feb 16, 2018 at 9:32 PM, Richard Marken rsmarken@gmail.com wrote:
[Rick Marken 2018-02-16_18:30:16]
Bruce Nevin (2018-02-16_12:27:28 ET)–
RM: It looks like you are controlling a program because you are keeping the program happening, protected from disturbances. I don’t know why you want to imagine that you are not controlling a program. But if you are keeping the program “on target” more than 80% of the time then you are controlling a program.Â
BN: Well, yes. I was controlling a perception of a program as means of controlling a perception of succeeding at your demo. But what program? And did I continue controlling the program after I figured out what to do?
BN: To control a perception of succeeding at your demo I attempted to follow the instructions. The instructions describe what your program is doing, and then say if it starts doing something else press the spacebar to get it to resume executing that program. The program is described as
“if the shape is circle, the next color is blue; else, the next color is red”
BN: The interpretation is that I perceive when the program is running and when it is not. I found it not a simple matter to recognize when the program is running. I have described how I used verbal ‘training wheels’, saying “blue next” when I saw a circle. And then after a delay adding “red next” when I saw a square, so I was controlling both “circle-blue” and “square-red”.Â
BN: As my performance improved, it seemed to me that I was controlling sequence perceptions: a circle-blue sequence and a square-red sequence. A bit tricky, controlling two contra-related sequences concurrently.
BN: You reply “No, those aren’t sequences, those are if-then contingencies. You just left of the ‘if’.” But then you say it’s just a matter of our using different words, drawing an analogy to the coin game illustration in B:CP.
Then you would see that we are in the position of the E and S in that game. The E concludes that S is controlling for the letter N while S protests that he was controlling for the letter Z. As Bill says “If they are both word oriented types, E and S may argue about whose definition is the “right” one, forgetting that E has discovered what S was in fact controlling, whatever either of them like to call it”. The demo shows that you are controlling for what I call a program: if circle then blue else red. You like to call it a sequence. But whatever you call it, when the on target score for what I call “program” goes above 80% I know that you are controlling for “if circle then blue else red”.
BN: Z and N are the same configuration with a different angular rotation, and there are well-established and well-practiced input functions for both those configurations, and, obviously, both are at the same level of the perceptual hierarchy.Â
BN: Here’s another program that I should do the same thing:
color=red
while i<121 do
  if odd(random) then
  paint(circle, color)
  color=blue
  else
  paint(square, color)
  color=red
  fi
  i=i+1
done
BN: The only contingency is to determine whether to paint a circle or a square as determined by a random number generator. I don’t know which program is running in the computer. So implement that, use a random process to run first one and then the other, and see if you can perceive the program that is running.
BN: But I don’t need to generate the series of colored shapes, nor do I need to recognize what program is running. I only need to press the spacebar when I perceive either of two sequences:
a circle followed by a red shape
a square followed by a blue shape
BN: So that’s all I look for to exercise the demo. I just need those two sequence-level input functions. Then the whack-a-mole sequences are
circle-red-bang!
square-blue-bang!
BN: Sorry to be a bad student, but that’s what I’m actually doing. Having figured out how to work it, I’m neither perceiving nor controlling either of the above two programs.
RM: I really appreciate your comments. But before I try to answer them I would like to know what your point is. Is your point that we (or you) don’t control programs? Or is that that my demo doesn’t demonstrate control of programs very well? Or is it something else?
BestÂ
Rick
Â
–
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
On Thu, Feb 15, 2018 at 1:37 PM, Richard Marken rsmarken@gmail.com wrote:
[Rick Marken 2018-02-15_10:37:34]
Bruce Nevin (2018-02-15_07:45:47 ET) –
BN: Given an initial state (a red or blue square or circle in one of the four positions on the circle), the structure of the program that is running in the computer may perhaps be stated as follows:
Â
while i < runtime do
 If shape=circleÂ
then colornext=blue
 else colornext=red
 if even(getrandom)Â
then shape=square
 else shape=circle
 fi
 position=position+1
 paint shape at position
 i=i+1
done
Â
RM: Not quite but irrelevant to the results.Â
BN: To the observer with inside knowledge of the computer program, it certainly looks as though I am controlling a perception of the computer program, but I am not.
RM: It looks like you are controlling a program because you are keeping the program happening, protected from disturbances. I don’t know why you want to imagine that you are not controlling a program. But if you are keeping the program “on target” more than 80% of the time then you are controlling a program.Â
Â
BN: I believe the same behavior could result from controlling the following two sequences concurrently:
square then blue then press
circle then red then press
 RM: Those are “if-then” contingencies, not sequences; you just left off the “ifs” at the beginning.
BN: Whenever the input functions of either of these sequence-control systems successfully controls the requisite perceptual inputs for the first two terms of the sequence it proceeds to control the third term of the sequence. Difficulty controlling two disparate sequences simultaneously will surely hamper performance at higher rates of changing the display.
BN: I have not invested the practice time to test this belief.Â
RM: Good decision; it would be a terrible waste of time. I think it would be much better to spend the time reading the section of B:CP on the coin game (pp. 236-238 in the 2nd edition). Then you would see that we are in the position of the E and S in that game. The E concludes that S is controlling for the letter N while S protests that he was controlling for the letter Z. As Bill says “If they are both word oriented types, E and S may argue about whose definition is the “right” one, forgetting that E has discovered what S was in fact controlling, whatever either of them like to call it”. The demo shows that you are controlling for what I call a program: if circle then blue else red. You like to call it a sequence. But whatever you call it, when the on target score for what I call “program” goes above 80% I know that you are controlling for “if circle then blue else red”.
Best
Rick
Â
–
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
On Mon, Feb 12, 2018 at 6:04 PM, Richard Marken rsmarken@gmail.com wrote:
[Rick Marken 2018-02-12_15:04:18]
Eetu Pikkarainen (2018-02-12_11:46:55 UTC)–
Â
EP: Sorry, I seem not to have patience enough - 1,5 min is too long for meÂ
 RM: I understand. I do think it might be worth it to try, though.Â
EP: But more seriously I feel that the concept of sequence is problematic for me here. The program can continue the whole time but isn’t the sequence always only three events long.
RM: Yes.Â
Â
EP: So to
control these sequences I must keep counting 1,2,3 again and again.
RM: Yes, I do that too; or I say “small, medium, large, small, medium…”. There is a lot of cognitive activity going on when I control the sequence or the program perception; apparently this is part of process involved in perceiving sequences and programs. It may be a feature rather than a bug.
Â
EP: If I miss one event from counting then the sequence changes to
“medium”, “large”, “small”.
RM: Yes, but I do think we are able to perceive (and thus control for) repeating sequences even when there is no demarcation of the beginning and end of the sequence. This certainly happens with speech sequences, where there is often no acoustical marker (like a pause) signaling the beginning and end of phoneme sequences that make up words. I can control the sequence pretty easily in that demo; with some practice I’m sure you could learn to do it too.Â
EP: And I confuse the counting every time I must press space bar. Should there be some mark for beginning of every sequence?
RM: I don’t think so. Try controlling the sequence at the “Slow” rate before moving on to the “Medium” rate. Don’t give up. The pay-off will be experiencing two of the more “cognitive” types of perceptual variable that Bill Powers discussed in B:CP: sequence and program.Â
Best
Rick
Â
Â
Eetu
Â
From: Richard Marken [mailto:rsmarken@gmail.com]
Sent: Sunday, February 11, 2018 4:09 AM
To: csgnet@lists.illinois.edu
Subject: Re: Controlling Sequences and Programs
Â
[Rick Marken 2018-02-10_18:08:32]
Â
RM: Based on some of Bruce’s comments I’ve put up what I think is an improved version of the sequence/program control demo at:Â
Â
http://www.mindreadings.com/ControlDemo/ProgramControl.html
RM: The easiest change was to have the shapes appear sequentially in different random positions along the central horizontal axis of the display. Somewhat more difficult was to tidy up some things that weren’t quite right. One was making sure that the computation
of the measure of control (proportion of trial on target) gave the right result. Now it should be true that if you are unable to control either the sequence or the program (as should be the case at the “Fast” display rate) the proportion of the trial on target
for both variables should be ~0.5.
Â
RM: If you can control the sequence but not the program (as should be the case at the “Medium” display rate) then the
proportion of the trial that is on target when you are controlling the sequence should be >0.8 for the sequence and ~0.5 for the program; and the proportion of the trial that is on target when you are unsuccessfully controlling the program should by ~0.5 for
both the program and the sequence.
Â
RM: And if you can control both the sequence and the program (as should be the case at the “Slow” display rate) then the proportion
of the trial that is on target when you are controlling the sequence should be >0.8 for the sequence and ~0.5 for the program; and the proportion of the trial that is on target when you are controlling the program should be should be >0.8 for the program
and ~0.5 for the sequence.Â
Â
RM: And I’ve cut down the time per trial from 2 to 1 1/2 minutes. But it’s still a pretty demanding task.Â
Â
Eetu Pikkarainen (2018-02-09_07:02:19 UTC)Â
Â
EP: Bruce, I think that you (like I) are controlling both programs 1 and 2. Program 1 is the main perception to be controlled but in addition to that we control
also 2.
Â
RM: If that were the case, then wouldn’t it also be the case for the sequence? Sequence 1 would be the main one (small, medium, large) and sequence 2 would be (small, large, medium, press; small, medium large, don’t press).
I agree that program 1 is the main program controlled when you are controlling the program perception but I don’t think program 2 is actually a controlled perception. If you try to build a model of what you are doing in this demo I think you will find that
the bar press has to be the output that is used to keep the “main” program perception under control. Same with control of the sequence; the bar press is not part of the main sequence perception that is controlled; it is the means by which you keep that perception
under control.Â
Â
RM: But, again, the main point of this demo is to show what we mean in PCT when we hypothesize that a program (or sequence) of events (including one’s own actions) is a
type of perceptual variable that is controlled.Â
Â
Martin Taylor (2018.02.09.17.28)–
Â
RM: Did no one try this?
Â
MT: I did, briefly, but I figured it would take many hours of practice to allow my hierarchy to reorganize to the point where I would
be testing the hierarchic levels rather than conscious perception, and I didn’t think I had the time to spare. But it’s a neat idea. How long did it take you to reorganize so that you were no longer controlling conscious perceptions? If it wasn’t as long as
I imagine, I might try again.
Â
RM: I don’t think you have to worry about whether this is “testing hierarchic levels rather than conscious perception”. I consider this demo to be an extension of Bill’s demonstration of the control of different types
of variables that can be found in “STEP H: BEYOND TRACKING” in Adam Matic’s phenomenally good javascript rewrites of Bill’s original Pascal demonstrations of PCT at:
Â
http://www.pct-labs.com/tutorial1/index.html
Â
RM: In STEP H you are given the option of controlling variables that include the relative size of two objects, the orientation of a shape, the shape of an object, the pitch of a sound and the size of a number. My demo
adds to this list a sequence of shape sizes  and a program of contingencies between shapes and colors.Â
Â
RM: I don’t think it takes that long to learn to control these variables; you should be able to do whatever reorganization is necessary for reasonably successful control after two or three trials in the “Slow” condition
(which I would suggest that everyone start with first).Â
Â
RM: It would be great if a number of you could do this demo and report the result (in terms of the proportion of a trial that the controlled variable – sequence or program – is kept on target when you try to control
these variables at the different display rates: Fast, Medium and Slow. Then maybe we could start discussing what the results mean and how to build a model – or, even better, a robot --Â that can imitate the behavior we see in this demo.Â
Â
Best regards
Â
Rick
–
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
–
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