[From Rick Marken (2017.11.30.1930)]
Erling Jorgensen (2017.11.30 0735 EST)
>RM [Responding to Martin Taylor]: But your rephrasing of the issue as "controlling to avoid a particular reference value for the controlled perception as opposed to "controlling to approach a particular reference value for the controlled perception" shows that I did not understand what the "issue" was at all. Indeed, your framing of the issue in this way makes it difficult for me to respond in any way other than to say that this issue has nothing to do with PCT. "Controlling to avoid a particular reference value for a controlled perception" is not "control of avoidance" because it's not control.Â
EJ:Â Gosh, Rick, these pronouncements of yours are difficult to take.Â
RM: I'm sorry to hear that. I'm really just saying these "pronouncements" based on my understanding of PCT. Maybe you, like everyone else on CSGNet, understands PCT better than I do. It doesn't seem like it to me but that's just my opinion and I am definitely in the minority here.
Â
EJ: The Crowd demo has a form of collision avoidance built into it, although I forget just now how it is implemented.Â
RM: It's implemented as control of proximity to other agents. Each agent acts so as to maintain a particular reference value of proximity relative to the other agents. It is not implemented as "controlling to avoid a particular reference value for a controlled perception" because that statement implies that each agent acts to move the controlled perception -- proximity - away from the reference value, which is obviously not control.Â
EJ: Ship captains learn that if the angle of displacement (is that the correct term?) of an orthogonal moving ship is not changing, then they either veer away or slow down to pass behind the other ship. Air traffic controllers try to keep approaching aircraft X-distance apart, and/or at different elevations. There are any number of ways to implement avoidance, whether by controlling distances or displacement angles or one's own velocity. I cannot see that this has "nothing to do with PCT."Â
RM: I think it has everything to do with PCT. The captain and the air traffic controller are acting to control a perception of the relationship between one moving object and another. They are not "controlling to avoid a the reference value for that relationship perception" ; they are controlling for keeping that relationship at the reference value.Â
RM: Perhaps my answer to Martin's statement about what he thought was the "issue" was too terse so I'll try to give a more complete reply.Â
RM: "Avoidance" and "approach" are words used to describe certain behaviors, like those of the ship captain and the air traffic controller. Martin gave a long list of behaviors that could be described as "avoidance" behaviors. As you may or may not know, PCT explains behavior as the control of perception. So the explanation of all these "avoidance" behaviors is the same as the explanation of all behavior -- "approach" behavior, "interception" behavior, "classy" behavior, "violent" behavior, etc. Indeed, the central focus of research based on PCT is aimed at discovering the perceptual variables that are under control when we see an organism behaving in a particular way. But such investigations have to start with a guess about the perceptual variable(s) being controlled. So I will try to demonstrate what I mean by guessing at the perceptual variables that might be under control in some of Martin's examples of "avoidance" behavior. Martin's examples are preceded by a "•¢ ":
•Â Â avoiding bumping into anyone in a crowd:
RM: A higher level system controls a perception of "bumping" at a reference of zero. This system keeps the bumping perception at zero by setting a reference for a lower level system controlling distance to other agents at some non-zero value: CVs = bumping, distance
Â
•Â    avoiding falling into the old mineshaft in the field.
RM: A higher level system controls a perception of falling into holes in the ground at a reference of zero. This system keeps from falling into holes by setting a reference for a lower level system controlling distance to holes in the ground at some non-zero value: CVs = falling in holes, distance
Â
•Â Â avoiding falling over the balustrade on the sevventh floor balcony.
RM: A higher level system controls a perception of falling over edges at a reference of zero. This system keeps from falling over edges by setting a reference for a lower level system controlling center of gravity at "over one's feet": CVs = falling, center of gravity
Â
•Â Â avoiding seeing the wine glass too near the edge of the table.
RM: A higher level system controls a perception of glasses falling off tables at a reference of zero. This system keeps the glass from falling by setting a reference for a lower level system controlling the distance from the glass to edge at about 5 inches: CVs = falling, distance.
Â
•Â Â avoiding hearing fooreigners talking their disgusting language in the bus.
RM: A higher level system controls a perception of hearing the sounds of foreigners talking at a reference of zero. This system keeps the sound from the foreigners from being heard by setting a reference for a lower level system controlling the distance to foreigners at a distance that keeps the sounds level from the foreigners near zero: CVs = sound level, distance.
•  avoiding offending that person with whose policies I disagree.
RM: A higher level system controls a perception of offensiveness of talk with this person at a reference of zero. This system keeps the talk inoffensive by setting a reference for a lower level system controlling the conversational topic to "the weather" (CV = offensiveness, topic of conversation)
RM: I think that's enough to give an idea of how PCT might explain behavior that we see as "avoidance". Of course, to really explain it you have to do the research to see what perceptual variables people are actually controlling when you see them doing these things.Â
RM: By the way, to see how PCT explains an example of "approach" behavior in terms of controlled perceptions I recommend a paper called "Chasin' Choppers" (<https://www.dropbox.com/s/eymkj4bxuorpyuy/Chasin'Choppers.pdf?dl=0>https://www.dropbox.com/s/eymkj4bxuorpyuy/Chasin'Choppers.pdf?dl=0\) which contains a nice test of some control models of approaching toy helicopters. It would be pretty easy to turn these into models what "avoid" helicopters by changing the references for the variables controlled when "approaching" the helicopters.Â
BestÂ
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