[philip 2018.03.20]
I feel as though discussion of PCT have degenerated quite a bit since the definition of its pure state in 1960. I don’t think PCT is about robotics, or about defining the conditions where control is “good”, or even about the TCV. PCT is simply about the general mechanism developed in Powers 1960. I don’t think any of the research on PCT today helps develop the ideas in the 1960 paper.
A quote:
We do not need to go through the tedious process of the “test of the significant variable” to obtain every bit of information we are going to accumulate. Both the human subject and the investigator are presumably similar creatures, and the investigator can often find short-cuts by an introspective analysis of his own perceptual methods. The investigator can hope to discover those variables which in his experience are “self-evident” classes, that is, which to his knowledge and belief are the forms in which he must perceive and has has always perceived his universe. The results of this introspection, in the form of classes of variables which should be significant to other human control systems, can also be subjected to the test of the significant variable. (Powers 1960, part 1)
In his paper, Powers says that work needs to be done in discovering the “classes” of perceptual variables (i.e. the sequences of static configurations) which are controlled in general.
What class of perceptual variables is controlled in drawing the ellipse? Here it is: We need
two foci, so the pen goes past the first focus and stops. Then it goes around the focus, past the second focus, and stops again. Then it goes back to where it started and stops a third time. This is a three term sequence of static first order configurations.
As Adam pointed out, nobody is controlling the curvature or velocity. If there is some relationship between the curvature and velocity of drawing, then it has nothing to do with PCT, because these variables are not controlled as a class of precepts. There is no behavioral illusion. The only illusion Powers mentions in his 1960 paper are the illusions inherent in the geometry of perception.
None of the questions in the original paper are topics of research in PCT today. Instead, I see people reinventing the robot, adding 6 new, redundant classes of perception, repeatedly doing experiments where the LOV is the significant variable, etc. I get it, the spinal cord doesn’t have an axon which represents “F = ma”. That doesn’t mean you can say robots are not built properly.
I don’t know everything about what you have been researching for 60 years, but nothing I have read is better than the original 1960 paper. Maybe you should just get people to read the 1960 paper instead of writing all these new books.
PCT is a theory of abstraction. It’s focus is on discovering the transforms which occur between orders of perception. It’s focus is not on discovering the mechanisms of motor control. Powers published his 1960 paper in the journal Perceptual and Motor Skills! For heaven’s sake, he already said in the paper everything one needs to know about motor skills. He says the proprioceptive feedback-connections in the first-order spinal loops divide before going to the central systems. That’s the first order. But you can’t just start talking about controlling a perception of scientific honesty when all you can do so far is draw an ellipse using a sequence of first order perceptions.
And I now I have a question for whoever read my thoughts:
How is it that the feedback function went from being at the input-output boundary in 1960 to being totally in the environment in LCS3?