What's an atenfel (was Re: Dynamic Visual Robot Arm Control)

CSGnet Archive CSG2016
1

[Martin Taylor 2016.07.04.10.48]

[From Rupert Young (2016.07.03 23.00)]

...
What's an atenfel?

Rupert

You can have a short and a moderately long answer here, or if you keep CSGnet postings, you can go back to my initial tutorial at [Martin Taylor 2016.06.07.16.56].

The short answer is that an atenfel (ATomic ENviromental Feedback ELement or ATomic ENvironmental FEedback Link -- take your pick) is any part of the environmental feedback path between the output of a control unit and the input to its perceptual function that can be taken as a unit.

Moderately long answer:

Often we talk about a physical object, such as a wire, as an atenfel, when really it is a property of the wire such as its conductivity that matters, not its shininess, diameter or whether it is copper, silver, or aluminum. Because it transmits a signal from one end to the other and that signal is a variable in the control loop, it is an atenfel. Even though it is in practice much quicker to talk about the object as atenfel, one must always remember that objects have an uncountable number of properties, of which only one or a few form part of the feedback path of the controlled perception with which we are concerned.

The properties of the environment that contribute to the controlled perception (what Rick calls the "controlled quantity despite it's not being controlled, and I call the CEV = Complex Environmental Variable) constitute a necessary atenfel in any control loop. In other words, the CEV is an atenfel, because without it, the environmental feedback path doesn't exist. The path from the output to the CEV and the path from the CEV to the perceptual function each consist of at least one atenfel, but may consist of many). Usually we are interested in only one of the many, as in the following examples.

The concept of atenfel is not very useful when we consider a single loop or a part of the hierarchy with fixed linkage weights. It becomes useful when a part of the environmental feedback path may change, perhaps because of uncontrolled physical effects (such as a bridge washing away in a flood), perhaps because the atenfel is the CEV of a controlled perception (laying a plank across a brook in order to be able to control a perception of being on the other side with dry feet), or because one atenfel is substituted for another in control of a particular variable (using the plank rather than a nearby pedestrian bridge). The concept is then useful because the environmental path changes only in part, and the way the change affects the environmental feedback function could be important.

For example, using the plank rather than walking to, across, and from the bridge changes the time it takes to reduce the error in the controlled perception of location. Both bridge and plank provide a "brook crossability dryshod" property that is needed as an atenfel for controlling a perception to a reference value of being on the other side with dry feet -- using "refer to the object" shorthand, we would usually say that the bridge and plank are potential atenfels for control of this perception, and that one of them becomes an active atenfel when it is actually incorporated into the environmental feedback path of a controlled perception.

When Kent McClelland and I were trying to find an appropriate name for the atenfel concept, we were dealing with social stabilities, the effects of one person's control of something to provide a stability that assists another's control of some perception. Trivially, if you want to drive a stake into the ground, you need two hands on the maul, and the task is very difficult because the stake keeps falling over before you can hit it. If now I control my perception of the stake's spatial orientation with a reference value of "vertical", your control for perceiving it to be embedded vertically in the ground becomes easy.

The stability of the initial spatial orientation of the stake is the atenfel. My control of my perception of its orientation provides that stability, but you could provide it for yourself if, say, there were some stones you could pile around the stake before you start to pound it. The same stability is the same atenfel, but by using the stones it is your own perceptual control that produces it, rather than mine. It doesn't matter; the atefel is the same.

In his LCS IV chapter, Kent discusses a lot of these stabilities that ease control by other people, most of them not involving physical objects such as the stake. For example, a stable managerial style (?controlled perception of "self-as-perceived-by-others" in the manager?) helps the people being managed to control perceptions relating to the work being managed. In sport or war it is easier to beat an opponent with a stable strategic or tactical style than one "you never know what they will do next". Those stabilities are potential atenfels for the people being managed, and for you trying to win the game.

In CSGnet discussions, the interactions among control units seem almost always to concern conflicts induced by the effects of one control-unit's control on the CEV for a perception controlled by another control unit. By thinking of the component parts of environmental feedback paths rather than just the disturbance variable, we obtain a much wider and deeper appreciation of the possibiities of interacting control systems. Those component parts are "atenfels", potential or active.

Martin

2

[From Rick Marken (2016.07.04.1230)]

···

Martin Taylor (2016.07.04.10.48)–

RY: What’s an atenfel?

MT: The short answer is that an atenfel (ATomic ENviromental Feedback ELement or ATomic ENvironmental FEedback Link – take your pick) is any part of the environmental feedback path between the output of a control unit and the input to its perceptual function that can be taken as a unit.

MT: The properties of the environment that contribute to the controlled perception (what Rick calls the "controlled quantity despite it’s not being controlled, and I call the CEV = Complex Environmental Variable) constitute a necessary atenfel in any control loop.

RM: Again, I note that Powers consistently used the term “controlled quantity” to refer to the aspect of the environment that corresponds to the controlled perception. The controlled quantity, q.i, is the controlled perception, p, as seen from the perspective of an observer of the control system. For example, if the perception controlled by the control system is the area of a rectangle, p = h * w, then the controlled quantity is q.i = h * w ; the controlled quantity, q.i, and the controlled perception, p, are exactly the same variable seen from different perspectives. Therefore, it is incorrect to say that the controlled quantity, q.i, is not controlled when the controlled perception, p, is. The concept of a CEV is unnecessary at best and confusing at worst.

MT: The concept of atenfel is not very useful when we consider a single loop or a part of the hierarchy with fixed linkage weights. It becomes useful when a part of the environmental feedback path may change, perhaps because of uncontrolled physical effects

RM: I don’t understand why a new word helps here. The characteristics of the feedback path in a control loop are always changing; it’s the environment, after all. These variations are already taken into account in PCT as disturbance variables; environmental variations that can affect the state of the controlled variable. PCT diagrams typically show disturbances as having a direct effect on the controlled variable but Powers certainly recognized the fact that disturbances can also affect controlled variables indirectly through their effects on the feedback function. This is demonstrated in the demo I posted earlier:

http://www.pct-labs.com/tutorial1/index.html Step K. Control of Remote Effect

RM: In this demo the disturbances are the varying positions of the pulleys, which continuously change the gain of the feedback connection (what you are now calling an atenfel) from mouse movement to the controlled variable (the relationship between the target and the position of the top free end of the string). I think it’s good that you point out that disturbances can effect a controlled variable indirectly (via effects on the feedback function) but I’m not sure the introduction of a new term contributes to our understanding of how this influences the behavior of a control system.

MT: For example, using the plank rather than walking to, across, and from the bridge changes the time it takes to reduce the error in the controlled perception of location. Both bridge and plank provide a “brook crossability dryshod” property that is needed as an atenfel for controlling a perception to a reference value of being on the other side with dry feet – using “refer to the object” shorthand, we would usually say that the bridge and plank are potential atenfels for control of this perception, and that one of them becomes an active atenfel when it is actually incorporated into the environmental feedback path of a controlled perception.

RM: Looking at this example in terms of PCT I would say that what the person does depends on what perceptions she is controlling for. Assuming she just wants to get across the stream dryshod and do it without leaving the trail then she’ll control for going over the bridge; if she is in a hurry (controlling for getting back quickly) she’ll take the plank. If she is controlling for both remaining on the trail and getting back quickly then she will be in a conflict. The bridge and plank are not atenfels in PCT; they are states of a perceptual variable (connection between on side of stream and the other, a relationship perception) that can be controlled to achieve the higher level goal of crossing the river. The feedback functions are the physical characteristics of the bridge and plank that affect how one’s outputs (steps) related to the controlled perceptions – the perception of getting across the bridge or plank, depending on which is taken.

MT: When Kent McClelland and I were trying to find an appropriate name for the atenfel concept, we were dealing with social stabilities, the effects of one person’s control of something to provide a stability that assists another’s control of some perception.

RM: I hope there were some examples of PCT models of actual social interactions so that people can see exactly how PCT can explain social interactions. This was done quite nicely by Tom Bourbon in the paper he wrote for the volume of American Behavioral Scientist that I edited: Bourbon, T. (1990) Invitation to the Dance: Explaining the Variance when Control Systems Interact, American Behavioral Scientist, 34, 95-105. And he did it without the need to resort to an atenfel (or feedback function, for that matter).

MT: In CSGnet discussions, the interactions among control units seem almost always to concern conflicts induced by the effects of one control-unit’s control on the CEV for a perception controlled by another control unit. By thinking of the component parts of environmental feedback paths rather than just the disturbance variable, we obtain a much wider and deeper appreciation of the possibiities of interacting control systems. Those component parts are “atenfels”, potential or active.

RM: I agree that it would be worthwhile to study conflicts that are based on disturbances to the feedback functions of the conflicted systems. But I don’t see how the concept of atenfel helps.

Best

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

3

[From Rupert Young (2016.07.06 13.00)]

The short answer is that an atenfel (ATomic ENviromental Feedback ELement or ATomic ENvironmental FEedback Link -- take your pick) is any part of the environmental feedback path between the output of a control unit and the input to its perceptual function that can be taken as a unit.

Moderately long answer:

The properties of the environment that contribute to the controlled perception (what Rick calls the "controlled quantity despite it's not being controlled, and I call the CEV = Complex Environmental Variable) constitute a necessary atenfel in any control loop. In other words, the CEV is an atenfel, because without it, the environmental feedback path doesn't exist. The path from the output to the CEV and the path from the CEV to the perceptual function each consist of at least one atenfel, but may consist of many). Usually we are interested in only one of the many, as in the following examples.

So atenfels refer to things in the environment, in the feedback path. There may be properties associated with these things that affect the perceptions being controlled. But these properties are not properties of the control system. And the control system doesn't need to 'know' anything about them. They can all be bundled together as the feedback path?

Regards,
Rupert

···

On 04/07/2016 16:51, Martin Taylor wrote:

4

[Martin Taylor 2016.07.06.09.21]

[From Rupert Young (2016.07.06 13.00)]

The short answer is that an atenfel (ATomic ENviromental Feedback ELement or ATomic ENvironmental FEedback Link -- take your pick) is any part of the environmental feedback path between the output of a control unit and the input to its perceptual function that can be taken as a unit.

Moderately long answer:

The properties of the environment that contribute to the controlled perception (what Rick calls the "controlled quantity despite it's not being controlled, and I call the CEV = Complex Environmental Variable) constitute a necessary atenfel in any control loop. In other words, the CEV is an atenfel, because without it, the environmental feedback path doesn't exist. The path from the output to the CEV and the path from the CEV to the perceptual function each consist of at least one atenfel, but may consist of many). Usually we are interested in only one of the many, as in the following examples.

So atenfels refer to things in the environment, in the feedback path.

Yes, provided you remember that for any control unit above the lowest level the "environment" includes all lower level control systems that are affected by its output or that contribute to its perceptual input.

There may be properties associated with these things that affect the perceptions being controlled. But these properties are not properties of the control system.

They are properties of the control loop, but not of the elementary control unit (ECU) part of the control loop, consisting of Perceptual Input Function, Comparator, and Output Function. Whether to say "yes" or "no" depends on what you call "the control system".

And the control system doesn't need to 'know' anything about them.

A control unit has but one perceptual signal. That's all it "knows", so not only is there no need, there is no possibility.

They can all be bundled together as the feedback path?

Bundled by whom? Your wording suggests that the "bundle" is a perception somewhere, such as in an analyst of the control system, loop, or unit. If the analyst's interest is in the dynamics of the loop, then "bundling" might not be very efficient if the observable atenfels have easily described dynamic properties. If the analyst's interest is in experimentally using The Test for the Controlled Variable (TCV), then bundling might well be the right approach.

Maybe the perception is in someone being introduced to the idea of control. Then you, as teacher, probably wouldn't want to split out segments of the environmental feedback path, because it would complicate what you are trying to teach. Segmentation and the four types of interaction possibility among control systems can come later, probably much later.

Maybe the perception is in the mind of an inventor trying to find a way to improve the performance of a particular control loop. Then "bundling" would be quite the wrong thing to do, because improvement is usually best accomplished by substituting one segment with another constructed differently. For example, difficult control of the perceived location of a heavy rock can be made easy by the introduction of a wheelbarrow or a strong and willing helper into the part of the feedback loop between output and CEV.

When you are building robots, whether you want to be concerned with atenfels depends on what you are trying to achieve. If you want your robot to be able to switch modes (analogous to a person choosing to walk or bicycle) then you have to build in some kind of controller of perceptions based on separating our at least some atenfels. If you are building a robot that senses and controls only perceptions built entirely from the environment outside the robot's body, you probably won't be interested in segmenting any atenfels out of the feedback path. You always have three at least, but you won't care that you do.

If you are a sociologist, you may be interested in how some actions may assist or obstruct perceptual control by others, either deliberately or through side-effects (as opposed to interference by disturbing influences on the others' "controlled quantities" in the environment).

It comes down to what Rick often claims to be the only proper study in PCT: "what perception is being controlled". Some potential controlled perceptions are of the feedback path as a unit, some require segments of the path to be perceived and controlled separately, by independent control units either in the same or in a different "skin bag".

As a broad-brush generalization, the more control units are in perceptions you (observer/inventor/analyst/experimenter) want to control, the more likely it is that you will want to segment out some atenfels from some of their feedback paths. When you are concerned with only one feedback loop, you may often want to "bundle".

Martin

···

On 04/07/2016 16:51, Martin Taylor wrote:

5

[From Rick Marken (2016.07.08.1245)]

···

Rupert Young (2016.07.06 13.00)

RY: So atenfels refer to things in the environment, in the feedback path.
Martin Taylor (2016.07.06.09.21)–

MT: Yes, provided you remember that for any control unit above the lowest level the “environment” includes all lower level control systems that are affected by its output or that contribute to its perceptual input.

RY: So lower level control systems are atenfels along with components of the environmental feedback function. As I explained earlier, these are two very different beasts The environmental feedback function causes input while lower level control systems control input. And the environmental feedback function is in the environment while lower level control systems are in the organism. So it seems like it would be confusing to give these two different things the same name.

MT: As a broad-brush generalization, the more control units are in perceptions you (observer/inventor/analyst/experimenter) want to control, the more likely it is that you will want to segment out some atenfels from some of their feedback paths. When you are concerned with only one feedback loop, you may often want to “bundle”.

RM: I think I could get a better appreciation for the atenfel if you could show me how it is actually used in a simulation – a working model – of an example of some kind of behavior, such as a two person interaction, perhaps.

Best

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

6

[From Rupert Young (2016.07.08 21.30)]

(Martin Taylor 2016.07.06.09.21]

When you are building robots, whether you want to be concerned with atenfels depends on what you are trying to achieve. If you want your robot to be able to switch modes (analogous to a person choosing to walk or bicycle) then you have to build in some kind of controller of perceptions based on separating our at least some atenfels.

Are these, then, sub-goals; or lower control systems controlling other perceptual variables?

Rupert

7

[Martin Taylor 2016.07.08.17.03]

[From Rupert Young (2016.07.08 21.30)]

(Martin Taylor 2016.07.06.09.21]

When you are building robots, whether you want to be concerned with atenfels depends on what you are trying to achieve. If you want your robot to be able to switch modes (analogous to a person choosing to walk or bicycle) then you have to build in some kind of controller of perceptions based on separating our at least some atenfels.

Are these, then, sub-goals; or lower control systems controlling other perceptual variables?

Rupert

People have been known to refer to reference values supplied to lower-level control units as sub=goals. yes. I might, if I were talking to a non-PCT audience. In the present company I probably would call them supporting lower-level controlled perceptions.

But I don't see the relevance of the question to the example, which was to show why it is sometimes useful to separate out parts of an environmental feedback path without altering the perception being controlled. In my example switching one component for another would change the loop transport (literally) lag, but that's not always true, as it would not be if you replaced an old wire with fraying insulation with a shiny new one. The function of the circuit doesn't change when you switch one component for another that has the same properties within the circuit. The old wire and the new one are both atenfels for controlling a perception of the circuit function.

Here's another example. You are at the front door of a house to which you have been invited. You control for perceiving the door to be opened, so you control for a perception of hearing a bell ring inside, to do which you control a perception of pushing a particular button. You push the button and do not hear a bell ringing, and the door doesn't open so there's still error in the top-level perception. You switch the "bell sound perception" control for a "door knocker sound perception" control and use the available knocker to create a sound that you do hear. Soon the door opens. You switched the "bell sound" atenfel for the "knocker sound" atenfel for control of the "Door opening" perception. I guess you could call them subgoals if you want, one of them active, the other only potential at any one moment.

Martin

8

[From Rupert Young (2016.07.09 12.00)]

(Martin Taylor 2016.07.08.17.03]

[From Rupert Young (2016.07.08 21.30)]
Are these, then, sub-goals; or lower control systems controlling other perceptual variables?

People have been known to refer to reference values supplied to lower-level control units as sub=goals. yes. I might, if I were talking to a non-PCT audience. In the present company I probably would call them supporting lower-level controlled perceptions.

But I don't see the relevance of the question to the example, which was to show why it is sometimes useful to separate out parts of an environmental feedback path without altering the perception being controlled.

The relevance is that there is a difference between the feedback path that is inside the organism, and consists of other perceptual control systems, and the external feedback path which is the mechanics of the physical world. The use of a common term, atenfels, seems to be conflating two very different things. We already have terminology which separately covers these phenomena; sub-goals and the (external) feedback path.

Here's another example. You are at the front door of a house to which you have been invited. You control for perceiving the door to be opened, so you control for a perception of hearing a bell ring inside, to do which you control a perception of pushing a particular button. You push the button and do not hear a bell ringing, and the door doesn't open so there's still error in the top-level perception. You switch the "bell sound perception" control for a "door knocker sound perception" control and use the available knocker to create a sound that you do hear. Soon the door opens. You switched the "bell sound" atenfel for the "knocker sound" atenfel for control of the "Door opening" perception. I guess you could call them subgoals if you want, one of them active, the other only potential at any one moment.

Yes, they are, as you say, sub-goals; or "supporting lower-level controlled perceptions".

Regards,
Rupert

9

[Martin Taylor 2016.07.09.23.05]

[From Rupert Young (2016.07.09 12.00)]

(Martin Taylor 2016.07.08.17.03]

[From Rupert Young (2016.07.08 21.30)]
Are these, then, sub-goals; or lower control systems controlling other perceptual variables?

People have been known to refer to reference values supplied to lower-level control units as sub=goals. yes. I might, if I were talking to a non-PCT audience. In the present company I probably would call them supporting lower-level controlled perceptions.

But I don't see the relevance of the question to the example, which was to show why it is sometimes useful to separate out parts of an environmental feedback path without altering the perception being controlled.

The relevance is that there is a difference between the feedback path that is inside the organism, and consists of other perceptual control systems, and the external feedback path which is the mechanics of the physical world.

I don't see it that way at all, for two reasons. Consider: the hierarchy consists of an interconnected set of identically structured elementary control units (ECUs), each consisting of a perceptual input function (PIF), a comparator, and an output function (OF). The output function produces an effect on its environment, which for every ECU at a level above the lowest, consists of sending the output value as a contribution to the reference inputs of some comparators belonging to ECUs at the level below. The ECU of interest knows nothing of where its output goes. Only we analysts who know the circuitry can see that. Somehow or other, effects of changes in the output appear at the inputs to the PIF, again, as only we analysts know, from the perceptual outputs of lower level ECUs. All the ECU has to work with is it input and its output. All else happens in its environment, the connection being known as the environmental feedback path. We analysts know that the path isn't a simple straight wire connection, but has a lot of complications, including all of the lower level control units whose reference values are influenced by the ECU's output and all the lower level control units whose perceptual values contribute inputs to its PIF.

The second reason I don't see it as you do is that outside the skin-bag the "external" feedback path consists of a lot more than "the mechanics of the physical world" except in the very simplest of cases. There are a lot of independent control hierarchies in that external world, and some of them have effects on the feedback path other than contributing to the disturbance. The internal feedback path does not consist of "other perceptual control systems" because only half of each of those systems is internal to the skin bag. The other half is in the external world, just as is the case for the perceptual control loop of our focal ECU.

The use of a common term, atenfels, seems to be conflating two very different things. We already have terminology which separately covers these phenomena; sub-goals and the (external) feedback path.

Neither of which is a label for "a conceptually separable part of the feedback path". That's the only thing an atenfel is, a link in a chain. Of course different parts of the environmental feedback path, different atenfels, almost of necessity have different properties -- many more than two of them.

What you are saying is that we should have no word for a chain link because we already have a word for the whole chain, and the chain only works if all the links are present. An atenfel is not, and is not like, a subgoal. The control process that allows a subgoal to be achieved is an atenfel. Every control process functions inside and outside the skin bag.

A bridge provides an atenfel for perceiving oneself at a different place on the other side of a ravine, but it isn't the whole path. The road to the bridge from where you are is another part of the path, as is the road from the bridge to your target. All three are atenfels for controlling your perception of location with that reference value, as is the car in which you ride on those roads to get there (technically, the car-road complex is called a "molenfel" because neither is any use without the other being used in parallel, but I haven't wanted to go into networks of atenfels until the trivial part of the concept is clear. Bruce Nevin offered the example of a pen and paper as a molenfel for perceiving one's idea to be written down, neither pen nor paper being of any use for controlling that perception without the other). [You may have noticed that, for reasons explained earlier, I use the common shorthand form of referring to the object that has the atenfel property as though it were the actual link in the environmental feedback path.]

Just think of the chain, and of its links, and ask yourself whether an inspection of the chain for corroded or weak links is more conveniently done a link at a time or by pulling the chain until it breaks.

Martin

10

This is scary Martin... I think I'm beginning to see where the term might indeed be useful. I'm still going to vote (if permitted) that the term is quite ugly though! I think that I'm now getting the idea of how such a term could be useful in sociology. That would be, I think, for the analyst to examine possible existing atenfels for a particular situation (that needs improvement or resolution involving multiple persons) and then work on which one could improve the efforts (and goal achievement) for all of the involved persons.

Is this anywhere close?

···

On 07/09/2016 09:47 PM, Martin Taylor wrote:

[Martin Taylor 2016.07.09.23.05]

[From Rupert Young (2016.07.09 12.00)]

(Martin Taylor 2016.07.08.17.03]

[From Rupert Young (2016.07.08 21.30)]
Are these, then, sub-goals; or lower control systems controlling other perceptual variables?

People have been known to refer to reference values supplied to lower-level control units as sub=goals. yes. I might, if I were talking to a non-PCT audience. In the present company I probably would call them supporting lower-level controlled perceptions.

But I don't see the relevance of the question to the example, which was to show why it is sometimes useful to separate out parts of an environmental feedback path without altering the perception being controlled.

The relevance is that there is a difference between the feedback path that is inside the organism, and consists of other perceptual control systems, and the external feedback path which is the mechanics of the physical world.

I don't see it that way at all, for two reasons. Consider: the hierarchy consists of an interconnected set of identically structured elementary control units (ECUs), each consisting of a perceptual input function (PIF), a comparator, and an output function (OF). The output function produces an effect on its environment, which for every ECU at a level above the lowest, consists of sending the output value as a contribution to the reference inputs of some comparators belonging to ECUs at the level below. The ECU of interest knows nothing of where its output goes. Only we analysts who know the circuitry can see that. Somehow or other, effects of changes in the output appear at the inputs to the PIF, again, as only we analysts know, from the perceptual outputs of lower level ECUs. All the ECU has to work with is it input and its output. All else happens in its environment, the connection being known as the environmental feedback path. We analysts know that the path isn't a simple straight wire connection, but has a lot of complications, including all of the lower level control units whose reference values are influenced by the ECU's output and all the lower level control units whose perceptual values contribute inputs to its PIF.

The second reason I don't see it as you do is that outside the skin-bag the "external" feedback path consists of a lot more than "the mechanics of the physical world" except in the very simplest of cases. There are a lot of independent control hierarchies in that external world, and some of them have effects on the feedback path other than contributing to the disturbance. The internal feedback path does not consist of "other perceptual control systems" because only half of each of those systems is internal to the skin bag. The other half is in the external world, just as is the case for the perceptual control loop of our focal ECU.

The use of a common term, atenfels, seems to be conflating two very different things. We already have terminology which separately covers these phenomena; sub-goals and the (external) feedback path.

Neither of which is a label for "a conceptually separable part of the feedback path". That's the only thing an atenfel is, a link in a chain. Of course different parts of the environmental feedback path, different atenfels, almost of necessity have different properties -- many more than two of them.

What you are saying is that we should have no word for a chain link because we already have a word for the whole chain, and the chain only works if all the links are present. An atenfel is not, and is not like, a subgoal. The control process that allows a subgoal to be achieved is an atenfel. Every control process functions inside and outside the skin bag.

A bridge provides an atenfel for perceiving oneself at a different place on the other side of a ravine, but it isn't the whole path. The road to the bridge from where you are is another part of the path, as is the road from the bridge to your target. All three are atenfels for controlling your perception of location with that reference value, as is the car in which you ride on those roads to get there (technically, the car-road complex is called a "molenfel" because neither is any use without the other being used in parallel, but I haven't wanted to go into networks of atenfels until the trivial part of the concept is clear. Bruce Nevin offered the example of a pen and paper as a molenfel for perceiving one's idea to be written down, neither pen nor paper being of any use for controlling that perception without the other). [You may have noticed that, for reasons explained earlier, I use the common shorthand form of referring to the object that has the atenfel property as though it were the actual link in the environmental feedback path.]

Just think of the chain, and of its links, and ask yourself whether an inspection of the chain for corroded or weak links is more conveniently done a link at a time or by pulling the chain until it breaks.

Martin

11

[Martin Taylor 2016.07.10.16.44]

This is scary Martin... I think I'm beginning to see where the term might indeed be useful. I'm still going to vote (if permitted) that the term is quite ugly though!

Agreed! But Kent McClelland and I went through several possible names when we decided that a label for the concept was needed before we settled on "atenfel". The word does take some getting used to, but I am finding now that I use it quite naturally. The deciding factor for us was that the word lends itself to variation to handle situations that do come up. The most common variants (so far) are "molenfel" (MOLecular Environmental FEedback Link) (the pen-and-paper or car-and-road examples) and "atenex" (Atomic Environmental NEXus) a physical manifestation of which is a Swiss Army Knife that is designed to provide atenfels for a range of different controlled perceptions. There is also "molenex" but we haven't found much need for that so far. You could call a tool designed for one purpose, such as a saw, as a "desenfel" for control of a perception of separating something into two parts, following the same line of thought, but luckily we haven't ever used that word until this posting :slight_smile:

  I think that I'm now getting the idea of how such a term could be useful in sociology. That would be, I think, for the analyst to examine possible existing atenfels for a particular situation (that needs improvement or resolution involving multiple persons) and then work on which one could improve the efforts (and goal achievement) for all of the involved persons.

Is this anywhere close?

It's one possibility, but there are myriads of others. McClelland, at least in draft versions of his LCS IV chapter, has a section on the various work people do to create and maintain feedback paths for other people's control of all sorts of perceptions: "...without continual work a humanly structured environment begins to crumble over time..." In the physical environment, this is just the standard entropic decay, but McClelland applies it also to the intangible structure of organizations, entertainment, first responders, etc. etc. Every such feedback link, created or maintained so that other people (many of them unknown to the atenfel creator or maintainer) can more easily control some perception, is a part of a network of potential pathways converging and diverging through atenexes such as the abstraction we call "money". Much of this network is collectively controlled, no single control unit being responsible for the stability that makes it useful for controlling so many different perceptions in so many people.

Years ago, McClelland talked about a "gossamer network" of stabilities that allowed society to thrive. Now we have a more concrete way of treating that network.

...Even if we use ugly words to talk about it. If you can think of a beautiful set of words, I think Kent and I would be delighted. "Atenfel" has not yet seen the light of print, so it's not too late to change. Maybe there are possible combinatorial derivations from Greek or Latin or Old Norse or Sanskrit?

Martin

···

On 2016/07/10 12:09 AM, Bill Leach wrote:

On 07/09/2016 09:47 PM, Martin Taylor wrote:

[Martin Taylor 2016.07.09.23.05]

[From Rupert Young (2016.07.09 12.00)]

(Martin Taylor 2016.07.08.17.03]

[From Rupert Young (2016.07.08 21.30)]
Are these, then, sub-goals; or lower control systems controlling other perceptual variables?

People have been known to refer to reference values supplied to lower-level control units as sub=goals. yes. I might, if I were talking to a non-PCT audience. In the present company I probably would call them supporting lower-level controlled perceptions.

But I don't see the relevance of the question to the example, which was to show why it is sometimes useful to separate out parts of an environmental feedback path without altering the perception being controlled.

The relevance is that there is a difference between the feedback path that is inside the organism, and consists of other perceptual control systems, and the external feedback path which is the mechanics of the physical world.

I don't see it that way at all, for two reasons. Consider: the hierarchy consists of an interconnected set of identically structured elementary control units (ECUs), each consisting of a perceptual input function (PIF), a comparator, and an output function (OF). The output function produces an effect on its environment, which for every ECU at a level above the lowest, consists of sending the output value as a contribution to the reference inputs of some comparators belonging to ECUs at the level below. The ECU of interest knows nothing of where its output goes. Only we analysts who know the circuitry can see that. Somehow or other, effects of changes in the output appear at the inputs to the PIF, again, as only we analysts know, from the perceptual outputs of lower level ECUs. All the ECU has to work with is it input and its output. All else happens in its environment, the connection being known as the environmental feedback path. We analysts know that the path isn't a simple straight wire connection, but has a lot of complications, including all of the lower level control units whose reference values are influenced by the ECU's output and all the lower level control units whose perceptual values contribute inputs to its PIF.

The second reason I don't see it as you do is that outside the skin-bag the "external" feedback path consists of a lot more than "the mechanics of the physical world" except in the very simplest of cases. There are a lot of independent control hierarchies in that external world, and some of them have effects on the feedback path other than contributing to the disturbance. The internal feedback path does not consist of "other perceptual control systems" because only half of each of those systems is internal to the skin bag. The other half is in the external world, just as is the case for the perceptual control loop of our focal ECU.

The use of a common term, atenfels, seems to be conflating two very different things. We already have terminology which separately covers these phenomena; sub-goals and the (external) feedback path.

Neither of which is a label for "a conceptually separable part of the feedback path". That's the only thing an atenfel is, a link in a chain. Of course different parts of the environmental feedback path, different atenfels, almost of necessity have different properties -- many more than two of them.

What you are saying is that we should have no word for a chain link because we already have a word for the whole chain, and the chain only works if all the links are present. An atenfel is not, and is not like, a subgoal. The control process that allows a subgoal to be achieved is an atenfel. Every control process functions inside and outside the skin bag.

A bridge provides an atenfel for perceiving oneself at a different place on the other side of a ravine, but it isn't the whole path. The road to the bridge from where you are is another part of the path, as is the road from the bridge to your target. All three are atenfels for controlling your perception of location with that reference value, as is the car in which you ride on those roads to get there (technically, the car-road complex is called a "molenfel" because neither is any use without the other being used in parallel, but I haven't wanted to go into networks of atenfels until the trivial part of the concept is clear. Bruce Nevin offered the example of a pen and paper as a molenfel for perceiving one's idea to be written down, neither pen nor paper being of any use for controlling that perception without the other). [You may have noticed that, for reasons explained earlier, I use the common shorthand form of referring to the object that has the atenfel property as though it were the actual link in the environmental feedback path.]

Just think of the chain, and of its links, and ask yourself whether an inspection of the chain for corroded or weak links is more conveniently done a link at a time or by pulling the chain until it breaks.

Martin

12

[Kent McClelland (2016.07.10.1720)]

KM: I agree, too! Atenfel is not a pretty word, but I’m getting it used to it and have found it quite useful for describing the kinds of stabilities that we all rely on for making society work. All these good things have come to us thanks to perceptions that other people have maintained on our behalf, perceptions that we thereby don’t have to control ourselves.

KM: I’ve written a lot about conflict in the past, and that’s certainly an important part of the picture when you’re talking about society, but as a sociologist who prefers to look at the world through PCT glasses, I also want to be able to talk what gets produced when people cooperate with each other.

KM: Like Martin, I’d be happy to substitute some other word for atenfel, as long as it does the same job (or a better one) of helping me control my perception of how society works.

KM: PS: Thanks, Martin, for arguing my case for using atenfel so eloquently!

Kent

···

[Martin Taylor 2016.07.10.16.44]

On 2016/07/10 12:09 AM, Bill Leach wrote:

This is scary Martin... I think I'm beginning to see where the term might indeed be useful. I'm still going to vote (if permitted) that the term is quite ugly though!

Agreed! But Kent McClelland and I went through several possible names when we decided that a label for the concept was needed before we settled on "atenfel". The word does take some getting used to, but I am finding now that I use it quite naturally. The deciding factor for us was that the word lends itself to variation to handle situations that do come up. The most common variants (so far) are "molenfel" (MOLecular Environmental FEedback Link) (the pen-and-paper or car-and-road examples) and "atenex" (Atomic Environmental NEXus) a physical manifestation of which is a Swiss Army Knife that is designed to provide atenfels for a range of different controlled perceptions. There is also "molenex" but we haven't found much need for that so far. You could call a tool designed for one purpose, such as a saw, as a "desenfel" for control of a perception of separating something into two parts, following the same line of thought, but luckily we haven't ever used that word until this posting :slight_smile:

I think that I'm now getting the idea of how such a term could be useful in sociology. That would be, I think, for the analyst to examine possible existing atenfels for a particular situation (that needs improvement or resolution involving multiple persons) and then work on which one could improve the efforts (and goal achievement) for all of the involved persons.

Is this anywhere close?

It's one possibility, but there are myriads of others. McClelland, at least in draft versions of his LCS IV chapter, has a section on the various work people do to create and maintain feedback paths for other people's control of all sorts of perceptions: "...without continual work a humanly structured environment begins to crumble over time..." In the physical environment, this is just the standard entropic decay, but McClelland applies it also to the intangible structure of organizations, entertainment, first responders, etc. etc. Every such feedback link, created or maintained so that other people (many of them unknown to the atenfel creator or maintainer) can more easily control some perception, is a part of a network of potential pathways converging and diverging through atenexes such as the abstraction we call "money". Much of this network is collectively controlled, no single control unit being responsible for the stability that makes it useful for controlling so many different perceptions in so many people.

Years ago, McClelland talked about a "gossamer network" of stabilities that allowed society to thrive. Now we have a more concrete way of treating that network.

...Even if we use ugly words to talk about it. If you can think of a beautiful set of words, I think Kent and I would be delighted. "Atenfel" has not yet seen the light of print, so it's not too late to change. Maybe there are possible combinatorial derivations from Greek or Latin or Old Norse or Sanskrit?

Martin

On 07/09/2016 09:47 PM, Martin Taylor wrote:

[Martin Taylor 2016.07.09.23.05]

[From Rupert Young (2016.07.09 12.00)]

(Martin Taylor 2016.07.08.17.03]

[From Rupert Young (2016.07.08 21.30)]
Are these, then, sub-goals; or lower control systems controlling other perceptual variables?

People have been known to refer to reference values supplied to lower-level control units as sub=goals. yes. I might, if I were talking to a non-PCT audience. In the present company I probably would call them supporting lower-level controlled perceptions.

But I don't see the relevance of the question to the example, which was to show why it is sometimes useful to separate out parts of an environmental feedback path without altering the perception being controlled.

The relevance is that there is a difference between the feedback path that is inside the organism, and consists of other perceptual control systems, and the external feedback path which is the mechanics of the physical world.

I don't see it that way at all, for two reasons. Consider: the hierarchy consists of an interconnected set of identically structured elementary control units (ECUs), each consisting of a perceptual input function (PIF), a comparator, and an output function (OF). The output function produces an effect on its environment, which for every ECU at a level above the lowest, consists of sending the output value as a contribution to the reference inputs of some comparators belonging to ECUs at the level below. The ECU of interest knows nothing of where its output goes. Only we analysts who know the circuitry can see that. Somehow or other, effects of changes in the output appear at the inputs to the PIF, again, as only we analysts know, from the perceptual outputs of lower level ECUs. All the ECU has to work with is it input and its output. All else happens in its environment, the connection being known as the environmental feedback path. We analysts know that the path isn't a simple straight wire connection, but has a lot of complications, including all of the lower level control units whose reference values are influenced by the ECU's output and all the lower level control units whose perceptual values contribute inputs to its PIF.

The second reason I don't see it as you do is that outside the skin-bag the "external" feedback path consists of a lot more than "the mechanics of the physical world" except in the very simplest of cases. There are a lot of independent control hierarchies in that external world, and some of them have effects on the feedback path other than contributing to the disturbance. The internal feedback path does not consist of "other perceptual control systems" because only half of each of those systems is internal to the skin bag. The other half is in the external world, just as is the case for the perceptual control loop of our focal ECU.

The use of a common term, atenfels, seems to be conflating two very different things. We already have terminology which separately covers these phenomena; sub-goals and the (external) feedback path.

Neither of which is a label for "a conceptually separable part of the feedback path". That's the only thing an atenfel is, a link in a chain. Of course different parts of the environmental feedback path, different atenfels, almost of necessity have different properties -- many more than two of them.

What you are saying is that we should have no word for a chain link because we already have a word for the whole chain, and the chain only works if all the links are present. An atenfel is not, and is not like, a subgoal. The control process that allows a subgoal to be achieved is an atenfel. Every control process functions inside and outside the skin bag.

A bridge provides an atenfel for perceiving oneself at a different place on the other side of a ravine, but it isn't the whole path. The road to the bridge from where you are is another part of the path, as is the road from the bridge to your target. All three are atenfels for controlling your perception of location with that reference value, as is the car in which you ride on those roads to get there (technically, the car-road complex is called a "molenfel" because neither is any use without the other being used in parallel, but I haven't wanted to go into networks of atenfels until the trivial part of the concept is clear. Bruce Nevin offered the example of a pen and paper as a molenfel for perceiving one's idea to be written down, neither pen nor paper being of any use for controlling that perception without the other). [You may have noticed that, for reasons explained earlier, I use the common shorthand form of referring to the object that has the atenfel property as though it were the actual link in the environmental feedback path.]

Just think of the chain, and of its links, and ask yourself whether an inspection of the chain for corroded or weak links is more conveniently done a link at a time or by pulling the chain until it breaks.

Martin

13

[From Erling Jorgensen (2016.07.12 1215 EDT)]

I have not joined the discussion so far, because I only had a vague sense of what Kent & Martin wanted to use the word “atenfel” for. I agree on the ugliness factor, for the word itself (not the concept). And in fact, from the first discussions where the word came up, and continuing now into these more recent posts, I have steadily been substituting the word “feedlink” whenever I read “atenfel.”

I believe “feedlink” would be consistent with the intent of the concept, at least as described down through this series of attached posts. I also think it does a better job of reminding the reader what is being discussed, because it more naturally suggests ‘feedback linkage’ than having to remember & reconstruct the ‘At-En-Fe-L’ components of the constructed word. I have certainly noticed that mental impatience in myself at having to stop & recall what the word stands for. I think “feedlink” may just be more user-friendly to the reader, without (hopefully) losing a key connotation for the authors. Just my perceptual perspective.

All the best,

Erling

“McClelland, Kent” MCCLEL@Grinnell.EDU 7/10/2016 6:31 PM >>>
[Kent McClelland (2016.07.10.1720)]

KM: I agree, too! Atenfel is not a pretty word, but I’m getting it used to it and have found it quite useful for describing the kinds of stabilities that we all rely on for making society work. All these good things have come to us thanks to perceptions that other people have maintained on our behalf, perceptions that we thereby don’t have to control ourselves.

KM: I’ve written a lot about conflict in the past, and that’s certainly an important part of the picture when you’re talking about society, but as a sociologist who prefers to look at the world through PCT glasses, I also want to be able to talk what gets produced when people cooperate with each other.

KM: Like Martin, I’d be happy to substitute some other word for atenfel, as long as it does the same job (or a better one) of helping me control my perception of how society works.

KM: PS: Thanks, Martin, for arguing my case for using atenfel so eloquently!

Kent

[Martin Taylor 2016.07.10.16.44]

This is scary Martin… I think I’m beginning to see where the term might indeed be useful. I’m still going to vote (if permitted) that the term is quite ugly though!

Agreed! But Kent McClelland and I went through several possible names when we decided that a label for the concept was needed before we settled on “atenfel”. The word does take some getting used to, but I am finding now that I use it quite naturally. The deciding factor for us was that the word lends itself to variation to handle situations that do come up. The most common variants (so far) are “molenfel” (MOLecular Environmental FEedback Link) (the pen-and-paper or car-and-road examples) and “atenex” (Atomic Environmental NEXus) a physical manifestation of which is a Swiss Army Knife that is designed to provide atenfels for a range of different controlled perceptions. There is also “molenex” but we haven’t found much need for that so far. You could call a tool designed for one purpose, such as a saw, as a “desenfel” for control of a perception of separating something into two parts, following the same line of thought, but luckily we haven’t ever used that word until this posting :slight_smile:

I think that I’m now getting the idea of how such a term could be useful in sociology. That would be, I think, for the analyst to examine possible existing atenfels for a particular situation (that needs improvement or resolution involving multiple persons) and then work on which one could improve the efforts (and goal achievement) for all of the involved persons.

Is this anywhere close?

It’s one possibility, but there are myriads of others. McClelland, at least in draft versions of his LCS IV chapter, has a section on the various work people do to create and maintain feedback paths for other people’s control of all sorts of perceptions: “…without continual work a humanly structured environment begins to crumble over time…” In the physical environment, this is just the standard entropic decay, but McClelland applies it also to the intangible structure of organizations, entertainment, first responders, etc. etc. Every such feedback link, created or maintained so that other people (many of them unknown to the atenfel creator or maintainer) can more easily control some perception, is a part of a network of potential pathways converging and diverging through atenexes such as the abstraction we call “money”. Much of this network is collectively controlled, no single control unit being responsible for the stability that makes it useful for controlling so many different perceptions in so many people.

Years ago, McClelland talked about a “gossamer network” of stabilities that allowed society to thrive. Now we have a more concrete way of treating that network.

…Even if we use ugly words to talk about it. If you can think of a beautiful set of words, I think Kent and I would be delighted. “Atenfel” has not yet seen the light of print, so it’s not too late to change. Maybe there are possible combinatorial derivations from Greek or Latin or Old Norse or Sanskrit?

Martin

[Martin Taylor 2016.07.09.23.05]

[From Rupert Young (2016.07.09 12.00)]

(Martin Taylor 2016.07.08.17.03]

[From Rupert Young (2016.07.08 21.30)]
Are these, then, sub-goals; or lower control systems controlling other perceptual variables?

People have been known to refer to reference values supplied to lower-level control units as sub=goals. yes. I might, if I were talking to a non-PCT audience. In the present company I probably would call them supporting lower-level controlled perceptions.

But I don’t see the relevance of the question to the example, which was to show why it is sometimes useful to separate out parts of an environmental feedback path without altering the perception being controlled.

The relevance is that there is a difference between the feedback path that is inside the organism, and consists of other perceptual control systems, and the external feedback path which is the mechanics of the physical world.

I don’t see it that way at all, for two reasons. Consider: the hierarchy consists of an interconnected set of identically structured elementary control units (ECUs), each consisting of a perceptual input function (PIF), a comparator, and an output function (OF). The output function produces an effect on its environment, which for every ECU at a level above the lowest, consists of sending the output value as a contribution to the reference inputs of some comparators belonging to ECUs at the level below. The ECU of interest knows nothing of where its output goes. Only we analysts who know the circuitry can see that. Somehow or other, effects of changes in the output appear at the inputs to the PIF, again, as only we analysts know, from the perceptual outputs of lower level ECUs. All the ECU has to work with is it input and its output. All else happens in its environment, the connection being known as the environmental feedback path. We analysts know that the path isn’t a simple straight wire connection, but has a lot of complications, including all of the lower level control units whose reference values are influenced by the ECU’s output and all the lower level control units whose perceptual values contribute inputs to its PIF.

The second reason I don’t see it as you do is that outside the skin-bag the “external” feedback path consists of a lot more than “the mechanics of the physical world” except in the very simplest of cases. There are a lot of independent control hierarchies in that external world, and some of them have effects on the feedback path other than contributing to the disturbance. The internal feedback path does not consist of “other perceptual control systems” because only half of each of those systems is internal to the skin bag. The other half is in the external world, just as is the case for the perceptual control loop of our focal ECU.

The use of a common term, atenfels, seems to be conflating two very different things. We already have terminology which separately covers these phenomena; sub-goals and the (external) feedback path.

Neither of which is a label for “a conceptually separable part of the feedback path”. That’s the only thing an atenfel is, a link in a chain. Of course different parts of the environmental feedback path, different atenfels, almost of necessity have different properties – many more than two of them.

What you are saying is that we should have no word for a chain link because we already have a word for the whole chain, and the chain only works if all the links are present. An atenfel is not, and is not like, a subgoal. The control process that allows a subgoal to be achieved is an atenfel. Every control process functions inside and outside the skin bag.

A bridge provides an atenfel for perceiving oneself at a different place on the other side of a ravine, but it isn’t the whole path. The road to the bridge from where you are is another part of the path, as is the road from the bridge to your target. All three are atenfels for controlling your perception of location with that reference value, as is the car in which you ride on those roads to get there (technically, the car-road complex is called a “molenfel” because neither is any use without the other being used in parallel, but I haven’t wanted to go into networks of atenfels until the trivial part of the concept is clear. Bruce Nevin offered the example of a pen and paper as a molenfel for perceiving one’s idea to be written down, neither pen nor paper being of any use for controlling that perception without the other). [You may have noticed that, for reasons explained earlier, I use the common shorthand form of referring to the object that has the atenfel property as though it were the actual link in the environmental feedback path.]

Just think of the chain, and of its links, and ask yourself whether an inspection of the chain for corroded or weak links is more conveniently done a link at a time or by pulling the chain until it breaks.

Martin

  NOTICE: This e-mail communication (including any attachments) is CONFIDENTIAL and the materials contained herein are PRIVILEGED and intended only for disclosure to or use by the person(s) listed above. If you are neither the intended recipient(s), nor a person responsible for the delivery of this communication to the intended recipient(s), you are hereby notified that any retention, dissemination, distribution or copying of this communication is strictly prohibited. If you have received this communication in error, please notify me immediately by using the "reply" feature or by calling me at the number listed above, and then immediately delete this message and all attachments from your computer. Thank you.
···

On 2016/07/10 12:09 AM, Bill Leach wrote:

On 07/09/2016 09:47 PM, Martin Taylor wrote:

14

[From Rupert Young (2016.07.13 20.45)]

(Martin Taylor 2016.07.09.23.05]

I don't see it that way at all, for two reasons. Consider: the hierarchy consists of an interconnected set of identically structured elementary control units (ECUs), each consisting of a perceptual input function (PIF), a comparator, and an output function (OF). The output function produces an effect on its environment, which for every ECU at a level above the lowest, consists of sending the output value as a contribution to the reference inputs of some comparators belonging to ECUs at the level below. The ECU of interest knows nothing of where its output goes. Only we analysts who know the circuitry can see that. Somehow or other, effects of changes in the output appear at the inputs to the PIF, again, as only we analysts know, from the perceptual outputs of lower level ECUs. All the ECU has to work with is it input and its output. All else happens in its environment, the connection being known as the environmental feedback path. We analysts know that the path isn't a simple straight wire connection, but has a lot of complications, including all of the lower level control units whose reference values are influenced by the ECU's output and all the lower level control units whose perceptual values contribute inputs to its PIF.

Not sure of the relevance of this to the difference between elements, internal to the system, that are are control systems, and those that are external and aren't.

The second reason I don't see it as you do is that outside the skin-bag the "external" feedback path consists of a lot more than "the mechanics of the physical world" except in the very simplest of cases. There are a lot of independent control hierarchies in that external world, and some of them have effects on the feedback path other than contributing to the disturbance. The internal feedback path does not consist of "other perceptual control systems" because only half of each of those systems is internal to the skin bag. The other half is in the external world, just as is the case for the perceptual control loop of our focal ECU.

How can systems on the internal feedback path be external?

The use of a common term, atenfels, seems to be conflating two very different things. We already have terminology which separately covers these phenomena; sub-goals and the (external) feedback path.

Neither of which is a label for "a conceptually separable part of the feedback path". That's the only thing an atenfel is, a link in a chain. Of course different parts of the environmental feedback path, different atenfels, almost of necessity have different properties -- many more than two of them.

How about just saying feedback element?

What you are saying is that we should have no word for a chain link because we already have a word for the whole chain, and the chain only works if all the links are present. An atenfel is not, and is not like, a subgoal. The control process that allows a subgoal to be achieved is an atenfel. Every control process functions inside and outside the skin bag.

How can you have a control process outside of the skin bag (that is not part of another skin bag)?

A bridge provides an atenfel for perceiving oneself at a different place on the other side of a ravine, but it isn't the whole path.

Sounds a bit like affordances; as if inanimate objects have some sort of active causal power.

Regards,
Rupert