in the text….
···
From: Adam Matic [mailto:adam.matic@gmail.com]
Sent: Wednesday, September 07, 2016 3:56 PM
To: csgnet@lists.illinois.edu
Subject: Re: Bill Powers on applying PCT
[From Adam Matic 7.9.2016]
I think I see the problem of ‘what questions to ask’ a bit more straight-forward then how to answer those questions.
There is the question of ‘what the animal is doing’ or ‘how to name this behavior’.
Then there is ‘how it works’ as a description of underlying processes and interactions with the environment.
If we add the developmental and evolutionary perspectives, it is almost like Tinbergen’s four questions.
For the first question, if we assume that behavior is control,
HB : Sorry the assumption is wrong. Behavior is not control. But organism controls….Control of behavior is Ricks’ inveention, which I interpreted as »perceptual illusion«.
Behavior is not control and it’s not controlling any »controlled variable« in environment and it’s not controlling input. In PCT »Perception is controlled«. You can see from Bill’s diagram that behavior is the result of »error« signal«, which is the result of »Control of perception« in comparator.
Bill P (LCS III). :
AM : we also assume that every animal has its own perceptual world (umwelt, merkwelt, subjective experience…) then we need to try to identify as accurately as possible what is being perceived and controlled by the animal.
HB : Right. What is being perceived and contolled by animal !!! Not controlled by behavior…
AM : A lot of these questions have been answered without PCT.
HB : Right. Physiology and biology did a lot of work.
AM : We know bats don’t use vision much, but still perceive depth and shapes from ‘sonars’, dogs probably live in a very smell-dominated reality, snakes see temperature changes… When we see a frog catching flies, we can know it is not really ‘catching flies’ since it cannot differentiate between flies and falling pieces of cardboard (catches both).
The TCV is one tool for such studies,
HB : Quite unsufficient tool… if we understandd it in Rick’s sense.
AM : but it also requires a fair bit of imagination, guesswork and careful experimentation.
HB : I agree…J
AM : We can’t know first-hand what the animal is experiencing, but we can try to gradually improve our guesses and models by taking both behavioral and physiological or neuroimaging data.
HB : I agree. Try to explain to Rick this…
AM : It seems that this idea of acknowledging subjectivity in the animal is not always present (legacy of behaviorism?) but there are many examples where it is present in some form, as there has been a lot of pushback to behaviorism.
HB : I agree
AM : What researchers do after this step of identifying perceptions, depends on the theoretical framework they assume is correct.
HB : Bill did use some physiological knowledge (see B:CP). It’s not just theoretical framework. There are some experiments behind.
AM : For the ‘how it works’ question, PCT has a relatively clear framework:
HB : I would say no. It’s quite unclear if we are talking about framework (see bellow). The principle of »how it works« is quite clear. But not when we start using the orgsnisms PCT framework. Dag for example is waiting for some solution to come in future. But it can be that he will not see the solution.
AM : ….if we are concerned with simple behaviors, often we can go with just one or two loops for start.
HB : Even simple behaviors can be the result of very complex perceptions controlled by organism.
AM : After that there are many possible directions - staying on the same level and adding perceptions and behaviors that happen simultaneously to the model; we can model learning and reorganization; we can go down a level and model perceptual functions explicitly; we can model motor/output loops explicitly, including muscle dynamics or environmental forces; or we can go up a level and attempt to model higher processes and varying reference functions. Or we can do group behavior…
HB : I’m not sure I understand….:). Maybe some life examples would help.
<
AM : I think Bill did something along those lines with the tracking and arm models. First there is a single loop model of tracking. Then there is tracking that includes a simple arm model in the output. Then there is a model of the arm with some muscle dynamics and a visual system that does target tracking. There is also a motor learning / reorganization model.
HB . Beside mechanical (robotic) use of PCT I’m sure that we can use PCT in some other sciences. Motor learning/reoganization model ??? Maybe some examples ?
AM : These types of research are also being done without PCT, there are various agent models with Markov chains, neural networks, control loops; robotic models with artificial sensors and motors; all attempting to replicate animal behaviors in various levels of detail. My assumption is that PCT models would be somehow better then alternatives, more biologically plausible or more accurate in replicating modeled behavior.
HB : I agree. But there is quite a way to be done to improve PCT to the level of clear use.
AM : As such, those models could serve as starting questions for the developmental and evolutionary perspective - how exactly those loops and behaviors develop and change trough time.
HB : Some first organisms (cells) didn’t have any outer loop (behavior). They just survived with internal control loops inside membrane.
AM : To me, it seems there is a lot of confusion about terms such as ‘function’ and ‘purpose’ of animal behaviors, and a framework with a hierarchy of purposes might bring some order and clarity into it.
HB : Purposive behavior can be also clear in animals with simple structure, no hierarchy. For example if you put E.Coli into suitable environment it will »tumble« in the place. If you put sugar into environment, E.Colli will stop tumbling after some time as it will start to proppel direclty to the sugar probably following the path of moleculs coming from the sugar. So »purpose« to survive seems to be the start of the evolution. What can be more clear about »function« and » purpose« of animal behavior ?
Best,
Boris
Best,
Adam
On Tue, Sep 6, 2016 at 3:30 AM, Richard Marken rsmarken@gmail.com wrote:
[From Rick Marken (2016.09.05.1830)]
Vyv Huddy(0907.05.09.2016)
VH: Reading the debates over the last two months has brought one quote to mind a number of times:-
LCS II p.173 “to realize the potential of control theory, we must apply it directly and not think of it as a means to finding new answers to old questions. The old questions are loaded with assumptions, and their answers are probably meaningless. We need new questions, and the basic question, I suggest, is how people and animals actually work. We have a believeable model to start with”.
VH: The trouble is that this isn’t followed up with a heuristic framework to help us formulate good questions, apart from going up levels. There isn’t much more in the manifesto chapter in LCS II, at least from I can glean. Can anyone point me to anything more detailed on formulating research questions from PCT?
RM: Great question, Vyv. The question of how to formulate research questions from a PCT perspective is one I’ve struggled with for over 30 years. I’m afraid that there is no road map describing how to do this, other than what Bill says in the quoted passage above, which I will paraphrase as “Forget the old questions about how people and animals work and start asking new ones based on PCT”. Actually, maybe the “Cybernetic Model for Research in Human Development” paper starting on p. 167 in LCS I is kind of road map for research on PCT. You might like to take a look at that.
RM: I’ve managed, after 20 years of thinking about it, to formulate a few research questions from a PCT perspective. For example, “What are people controlling when they intercept moving objects”? ; “What is the evidence that different types of perceptions are controlled at different levels of a control hierarchy”?; “What is the cause of objectively observed errors in controlling”? I’ve done some research to test these questions and published papers describing the results of this research. But none of these projects have turned into research programs. I think this is because I’m working on my own (without the help of really bright people, like some of those here on CSGNet, who prefer trying to answer the old questions;-) and also because I don’t have the resources to make that happen.
RM: But I think the main impediment to being able to formulate research questions from a PCT perspective is the opposition to PCT that Bill speaks of in the beginning of that paper (p. 171). Bill thought (at the CSG meeting back in 1986, when this paper was written) that we were “approaching a time when we will no longer have to concentrate on overcoming opposition” to PCT. Thirty years later that time has still not arrived. But there was never really much active opposition to PCT. I think the opposition to PCT mainly existed (and still exists) in the form of the behemoth that is the conventional life science establishment – the huge infrastructure of courses, educational programs, textbooks and scientific publications that are based on a causal model of “how people and animals work”.
RM: I think this establishment is an impediment to formulating research questions from a PCT perspective for several reasons. First, it defines what are considered the appropriate research question and how to go about studying them. So anyone who has been trained in this establishment will have trouble thinking of a research question that is not informed by it’s point of view, which differs considerably from the PCT point of view. Second, if you do come up with a research question from a PCT perspective it will not be of much interest to anyone in that establishment; so PCT research doesn’t have an audience among the researchers who you would like to have be your audience (or research colleagues). Third, if you go ahead and do the research anyway and write it up you’ll have a heck of a time trying to publish it since establishment reviewers will not like or understand it.
RM: I think Bill ended up continuing his work aimed at “showing what’s wrong with conventional ideas” about how people and animals work (as he says in LCS II, p. 171) because every time he came in contact with the life science establishment (in the form of members of that establishment asking him how PCT would explain this or that phenomenon or telling him that PCT couldn’t explain this or that phenomenon) he was forced to “show what’s wrong with conventional ideas” (as I’m doing now with the idea of the power law). Because of this he wasn’t able to do much of the research he would have liked to do that would have provided a model of how to do research based on PCT. I know that Bill did try to do some PCT research by insinuating it into some conventionally conceived research that was being done by other people. But none of this turned out to be a very good example of PCT research because the people he was doing it with were firmly entrenched in the establishment perspective and Bill couldn’t get the research done the way he wanted it done.
RM: But I think this is an important topic, Vyv, so if you have any general ideas for research aimed at understanding, from a PCT perspective, how people and animals work, let’s knock them around and see if we can come up with any ideas for PCT research that we (or someone) could actually do. Or that would at least let people see what PCT research would look like.
Best regards
Rick
Richard S. Marken
“The childhood of the human race is far from over. We have a long way to go before most people will understand that what they do for others is just as important to their well-being as what they do for themselves.” – William T. Powers