Analyzing feedback paths

[From Kent McClelland (2012.11.02.1250)]

As a sociologist I'm interested in looking at the environment as consisting of possible feedback paths for controlling different kinds of perceptions. I've done some thinking about how one might classify feedback paths, and I'd be happy to hear any corrections, additions, or suggestions:

Feedback paths provide the foundation for a sociological analysis from the PCT perspective, because social interaction takes place in a shared physical environment. Feedback paths can be analyzed on a number of dimensions:

  1. Exclusivity�Inclusivity. Is this feedback path the �private property� of one individual or is it open to use by multiple individuals? How big is the pool of potential users? How many individuals depend on it for controlling their perceptions?
  2. Flexibility of Purpose. How many different kinds of perceptions does this feedback path support? Is it single-purpose or multi-purpose? How many orders of perception can be supported by this feedback path? How many different perceptions at any given level does this feedback path support for the individual user?
  3. Bandwidth. How strong is this feedback channel for maintaining control of a given perception? Feedback bandwidth may be increased by multiplying the effects of physical actions that compensate for disturbances, magnifying the quantity or quality of information passed to perceptual channels, or protecting the channel from possible disturbances.
  4. Fixity�Evanescence. How nearly permanent or resistant to random physical forces is this feedback channel? Does it need to be reconstructed every time a given perception is controlled?
  5. Manipulability. How easily can this feedback channel be manipulated (changed in a controlled way) to meet the objectives of the user?

Hypothesis: An individual�s liberty and flexibility of action is increased when the individual has access to feedback paths that are exclusive, multi-purpose, high bandwidth, and fixed but highly manipulable. (Which must be why people fall in love with the hand-held electronic devices that satisfy most of those requirements.)

Kent

[From Rick Marken (2012.11.02.1615)]

Kent McClelland (2012.11.02.1250)]

KM: As a sociologist I’m interested in looking at the environment as consisting of possible feedback paths for controlling different kinds of perceptions. I’ve done some thinking about how one might classify feedback paths, and I’d be happy to hear any corrections, additions, or suggestions:

RM: It would help me if you gave some concrete examples of each of these types of feedback paths. For example, I’d like a real world example of an exclusive feedback path and a non-exclusive one so I can see the difference. Right now I wouldn’t know on what basis to conclude that a person is controlling something via an exclusive feedback path, a flexible one, a high bandwidth one, and so on. Once I understand this (if I can;-) I might be able to provide some suggestions.

Best

Rick

···

Feedback paths provide the foundation for a sociological analysis from the PCT perspective, because social interaction takes place in a shared physical environment. Feedback paths can be analyzed on a number of dimensions.

    1.      Exclusivity—Inclusivity. Is this feedback path the “private property” of one individual or is it open to use by multiple individuals? How big is the pool of potential users? How many individuals depend on it for controlling their perceptions?


    2.      Flexibility of Purpose. How many different kinds of perceptions does this feedback path support? Is it single-purpose or multi-purpose? How many orders of perception can be supported by this feedback path? How many different perceptions at any given level does this feedback path support for the individual user?


    3.      Bandwidth. How strong is this feedback channel for maintaining control of a given perception? Feedback bandwidth may be increased by multiplying the effects of physical actions that compensate for disturbances, magnifying the quantity or quality of information passed to perceptual channels, or protecting the channel from possible disturbances.


    4.      Fixity—Evanescence. How nearly permanent or resistant to random physical forces is this feedback channel? Does it need to be reconstructed every time a given perception is controlled?

    5.      Manipulability. How easily can this feedback channel be manipulated (changed in a controlled way) to meet the objectives of the user?

Hypothesis: An individual’s liberty and flexibility of action is increased when the individual has access to feedback paths that are exclusive, multi-purpose, high bandwidth, and fixed but highly manipulable. (Which must be why people fall in love with the hand-held electronic devices that satisfy most of those requirements.)

Kent


Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[Martin Taylor 2012.11.03.08.13 CET]

[From Kent McClelland (2012.11.02.1250)]

Seeing your post reminded me that before I left home I was working on an answer to your "more about gain" posting in which you asked about the "composite feedback loop". The algebra isn't complicated, but there are a lot of symbols and I didn't trust my answers so I didn't post it before I went to Europe. I intended to post it just now after seeing your name, including errors that probably persist, but I find that the diagrams are held on my desktop machine, not this one, so I can't post until I get home in a couple of weeks.

As a sociologist I'm interested in looking at the environment as consisting of possible feedback paths for controlling different kinds of perceptions. I've done some thinking about how one might classify feedback paths, and I'd be happy to hear any corrections, additions, or suggestions:

All of your points are excellent, except that they refer to "individuals", by which I assume individual animals. They seem mostly to apply to individual control units, though the final hypothesis deals with the collection of control units in an individual animal.

Feedback paths provide the foundation for a sociological analysis from the PCT perspective, because social interaction takes place in a shared physical environment. Feedback paths can be analyzed on a number of dimensions:

  1. Exclusivity�Inclusivity. Is this feedback path the �private property� of one individual or is it open to use by multiple individuals? How big is the pool of potential users? How many individuals depend on it for controlling their perceptions?

A feedback path that can be used only by one control unit is not susceptible to disturbance from the directly conflicting actions of others. It is, however, subject to disturbance from the side-effects of the actions of others (the situation I am addressing in my response to your "more about gain" posting. So it is possible that even if the feedback path is used only by the one control unit, its disturbance is influenced by all the other control units in the world. To get to the "private property" equivalent, you have to shield the feedback path. At the level of individuals, we have laws that effect the shielding. If I own something, and I say you can't do anything to influence it, there may be legal reasons why you shouldn't, so you probably won't. But laws just reduce the probability, and in any case it is usually hard for either of use to know whether the side effects of your legitimate activity will disturb my control unit.

  2. Flexibility of Purpose. How many different kinds of perceptions does this feedback path support? Is it single-purpose or multi-purpose? How many orders of perception can be supported by this feedback path? How many different perceptions at any given level does this feedback path support for the individual user?

You are talking about what I have called "environmental affordances". A stone could serve as a part of an ornamental display, a pounder for driving in a stake, a projectile for throwing, and so forth. However, when it is used for one of these purposes, it cannot be used for another. If only that stone is available and at some moment the individual is trying to control two of those perceptions, there is conflict. If there is a heavy hammer at hand, and more throwable stones, then this pretty one can continue to serve its function in your feedback loop for controlling your perception of the ornamental display, eliminating the conflict. So "Flexibility of Purpose" must be considered along with the ranges of means by which all these other purposes can be achieved. The more environmental affordances associated with a particular feedback path, the more opportunity for conflict (it's the direct opposite of your "privacy" state), and the more environmental affordances in the environment for a particular purpose, the more opportunity for avoiding conflict. We usually discuss this under the heading "degrees of freedom".

  3. Bandwidth. How strong is this feedback channel for maintaining control of a given perception? Feedback bandwidth may be increased by multiplying the effects of physical actions that compensate for disturbances, magnifying the quantity or quality of information passed to perceptual channels, or protecting the channel from possible disturbances.

"Bandwidth" has an engineering meaning relating to how fast unpredictable changes can be propagated through a channel. That's the meaning appropriate to PCT discussions. You mention several different properties that may be important, such as the range and precision of possible effects of the output upon the environmental variable affected by a disturbance, the speed and precision of the perceptual pathway from the environment to the perceptual signal value, the shielding I mentioned in respect to your first point. All of these do affect how well the control system can control over a wide range of disturbance values, or, using a language closer to yours, how reliable is the feedback path.

  4. Fixity�Evanescence. How nearly permanent or resistant to random physical forces is this feedback channel? Does it need to be reconstructed every time a given perception is controlled?

Here you get into a pervasive issue in PCT: the interactions between control systems that act on their environment to control their perceptions (all control systems) and those whose perceptions are the working components of other control systems. A control system doesn't know anything about its feedback path. All that happens if the feedback path is broken is that control fails. In much PCT discussion, if control persistently fails, "reorganization" happens, and that is a random process. This doesn't seem to me to be what happens in real life. If I use a bridge on my regular route, and it is closed for maintenance one day, I may cast about until I find an alternate route. This alternate route works to control the perceptions relating to getting where I am going, so why should I go back to using the bridge when I learn that the maintenance work is complete? There must be some other perception whose control affects part of the "going to that place" set of control systems. How that works is something that needs to be addressed within PCT.

  5. Manipulability. How easily can this feedback channel be manipulated (changed in a controlled way) to meet the objectives of the user?

This also deals with the question of how one control system can control perceptions relating to the performance of another. Who (which control system) does the manipulating, in control of what perception? It is not control system whose feedback channel is being manipulated. "The objectives of the user" form one or more controlled perceptions that in their output pathways incorporate lower-level control systems. Your question presumes that at least one of these lower level systems would work better if its feedback path were manipulated. Who does the manipulating and why?

Hypothesis: An individual�s liberty and flexibility of action is increased when the individual has access to feedback paths that are exclusive, multi-purpose, high bandwidth, and fixed but highly manipulable. (Which must be why people fall in love with the hand-held electronic devices that satisfy most of those requirements.)

Kent

Apart from the use of the word "bandwidth" to encompass speed, range, and precision, that hypothesis seems most reasonable.

The word "access" is important, as it implies not only that an outside observer could know that a particular feedback path is available for the individual to control a particular perception, but that the individual has reorganized to use it when appropriate (not conflicting with other purposes -- see the example of the stone, above.

"Access" very often means "money", often enough that the possibility of other means is ignored. Money can limit access for some purposes, just as it increases access for most. The old saw says "money can not buy happiness", but it can buy many possibilities for control. There used to be a cartoon series "Happiness is a ..." (e.g. warm hug). So what characteristic in PCT corresponds to "happiness"?

Martin

[From Kent McClelland (2012.11.03.1445)]

Rick Marken (2012.11.02.1615)

RM: It would help me if you gave some concrete examples of each of these types of feedback paths. For example, I'd like a real world example of an exclusive feedback path and a non-exclusive one so I can see the difference. Right now I wouldn't know on what basis to conclude that a person is controlling something via an exclusive feedback path, a flexible one, a high bandwidth one, and so on. Once I understand this (if I can;-) I _might_ be able to provide some suggestions.

KM: Sure, I could provide some examples.

1. Exclusive vs. Inclusive feedback paths: your private, fenced-in backyard garden vs. a national park; a single-seater ultralight plane vs. a jumbo jet that carries hundreds of people.

2. Flexible vs. inflexible feedback paths: a Swiss army knife vs. the cherry pitter device sitting in our kitchen cupboard (which only does one thing and that not too well); the latest smart phone vs. an old-fashioned rotary-dial phone. (Note that the smart phone permits control at a number of different perceptual levels, from taking a photo or communicating with another person to the level of providing a symbolic indication of the kind of person you are--relatively rich, with it, and so busy you need to stay in constant touch with others.)

3. High-bandwidth vs. low-bandwidth feedback paths: a telescope for looking at the moon vs. a box with a pinhole in it; a bulldozer vs. a shovel.

4. Fixed vs. Evanescent feedback paths: A message on a tombstone vs. a message written on sand at the beach; the old family silver service vs. plastic forks and spoons from McDonalds.

5. Manipulable vs. Unwieldy (I'm not quite sure what to call the other end of this scale): a personal computer vs. a doorbell; your private car vs. a jumbo jet (manipulable by the pilots, we hope, but not by you). This scale seems to overlap a lot with some of the other scales--exclusive-inclusive, flexible-inflexible, and fixed-evanescent--so I'm not sure if it's an independent dimension, but this is just my first cut at trying to come up with a scheme for looking at objects in the world around us from the point of view of where they fit into control loops.

Backing up, the reason I'm focusing on feedback paths is because it seems like people interested in PCT have already done a pretty good job of exploring the psychological workings of the control loops inside a person's body, looking at perceptual input functions, comparator functions, output functions, and the various signals that connect these functions. But what makes PCT approach distinctive is that we insist that it's also important to consider feedback functions. However, my sense of where we're at is that the PCT analysis of feedback functions is relatively undeveloped, other than paying close attention to the influence of disturbances.

Most social scientists--including most economists, political scientists, sociologists, and anthropologists--aren't particularly interested in the psychological details of how neural connections work. Instead, they are intensely interested in understanding and explaining empirically observable objects, patterns of human action, and events that occur outside the human body. From the point of view of PCT, it seems to me, objects, actions and events can all be analyzed in terms of feedback functions. In the context of control loops, objects, actions, and events serve to provide feedback paths for controlling perceptions. If we can focus on explaining objects, actions, and events from a PCT perspective, we'll be meeting social scientists more than halfway.

The examples I've given above are all examples of objects used as feedback paths. In economic lingo, I'm talking about goods, and I think we could make a similar analysis of services: the patterns of action that serve as feedback paths for controlling the perceptions people want to control. Events might be a little trickier to analyze, except that they often occur as a confluence of objects and actions. An election, for instance, is an event that brings together objects like buildings, voting machines, and TV sets, with actions like debates, news reporting, and people going to the polls. If we look at how all these various things serve as feedback paths for controlling the various participants' perceptions, we're on our way to a PCT analysis of this complex phenomenon.

Choosing one kind of feedback path instead of another for controlling a perception can be consequential in lots of ways, and I imagine you could come up with some examples of that yourself . . .

Best,

Kent

Kent

···

On Nov 2, 2012, at 6:14 PM, Richard Marken wrote:

[From Rick Marken (2012.11.02.1615)]

Kent McClelland (2012.11.02.1250)]

KM: As a sociologist I'm interested in looking at the environment as consisting of possible feedback paths for controlling different kinds of perceptions. I've done some thinking about how one might classify feedback paths, and I'd be happy to hear any corrections, additions, or suggestions:

RM: It would help me if you gave some concrete examples of each of these types of feedback paths. For example, I'd like a real world example of an exclusive feedback path and a non-exclusive one so I can see the difference. Right now I wouldn't know on what basis to conclude that a person is controlling something via an exclusive feedback path, a flexible one, a high bandwidth one, and so on. Once I understand this (if I can;-) I _might_ be able to provide some suggestions.

Best

Rick

Feedback paths provide the foundation for a sociological analysis from the PCT perspective, because social interaction takes place in a shared physical environment. Feedback paths can be analyzed on a number of dimensions.

        1. Exclusivity�Inclusivity. Is this feedback path the �private property� of one individual or is it open to use by multiple individuals? How big is the pool of potential users? How many individuals depend on it for controlling their perceptions?
        2. Flexibility of Purpose. How many different kinds of perceptions does this feedback path support? Is it single-purpose or multi-purpose? How many orders of perception can be supported by this feedback path? How many different perceptions at any given level does this feedback path support for the individual user?
        3. Bandwidth. How strong is this feedback channel for maintaining control of a given perception? Feedback bandwidth may be increased by multiplying the effects of physical actions that compensate for disturbances, magnifying the quantity or quality of information passed to perceptual channels, or protecting the channel from possible disturbances.
        4. Fixity�Evanescence. How nearly permanent or resistant to random physical forces is this feedback channel? Does it need to be reconstructed every time a given perception is controlled?
        5. Manipulability. How easily can this feedback channel be manipulated (changed in a controlled way) to meet the objectives of the user?

Hypothesis: An individual�s liberty and flexibility of action is increased when the individual has access to feedback paths that are exclusive, multi-purpose, high bandwidth, and fixed but highly manipulable. (Which must be why people fall in love with the hand-held electronic devices that satisfy most of those requirements.)

Kent

--
Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[From Rick Marken (2012.11.04.0840)]

Kent McClelland (2012.11.03.1445)

Rick Marken (2012.11.02.1615)

RM: It would help me if you gave some concrete examples of each of these types of feedback paths. …

KM: Sure, I could provide some examples.

  1. Exclusive vs. Inclusive feedback paths: your private, fenced-in backyard garden vs. a national park; a single-seater ultralight plane vs. a jumbo jet that carries hundreds of people.

RM: Thanks, Kent. this is a good start. But my understanding is that a feedback path is a connection between system outputs and a controlled variable. So in order to get a better understanding of your feedback paths I have to know what variables you think are being controlled and the means people use to control them. Once I know these things – controlled variables and the means available for controlling them – I think I’ll be able to understand better your classification of the feedback paths connecting them.

KM: Backing up, the reason I’m focusing on feedback paths is because it seems like people interested in PCT have already done a pretty good job of exploring the psychological workings of the control loops inside a person’s body, looking at perceptual input functions, comparator functions, output functions, and the various signals that connect these functions. But what makes PCT approach distinctive is that we insist that it’s also important to consider feedback functions. However, my sense of where we’re at is that the PCT analysis of feedback functions is relatively undeveloped, other than paying close attention to the influence of disturbances.

RM: I think what makes PCT distinctive is the emphasis on the fact that behavior is organized around the control of perceptual variables (controlled variables) and that understanding behavior is really a matter of learning what these variables are and how they are controlled. I think one has to do research aimed at the discovery of controlled variables before it makes any sense to analyze feedback functions since (as I said above) feedback functions are connections between system outputs and controlled variables. If you don’t know what variables are being controlled you can’t possibly know which outputs are affecting them or how they are affecting them (the nature of the feedback function).

KM: Most social scientists–including most economists, political scientists, sociologists, and anthropologists–aren’t particularly interested in the psychological details of how neural connections work. Instead, they are intensely interested in understanding and explaining empirically observable objects, patterns of human action, and events that occur outside the human body.

RM: I agree. And I think that what PCT shows is that in order to understand these things social scientists have to understand what variables people are controlling, how they control them (the lower order means used) and why (the higher order reasons for controlling the variable at particular levels).

KM: From the point of view of PCT, it seems to me, objects, actions and events can all be analyzed in terms of feedback functions.

RM: Again, not before you find out what variables are being controlled and the outputs used to control them. Once you know those things then the nature of feedback functions would, indeed, contribute to our understanding of observable aspects of human behavior.

Best

Rick

···

In the context of control loops, objects, actions, and events serve to provide feedback paths for controlling perceptions. If we can focus on explaining objects, actions, and events from a PCT perspective, we’ll be meeting social scientists more than halfway.

The examples I’ve given above are all examples of objects used as feedback paths. In economic lingo, I’m talking about goods, and I think we could make a similar analysis of services: the patterns of action that serve as feedback paths for controlling the perceptions people want to control. Events might be a little trickier to analyze, except that they often occur as a confluence of objects and actions. An election, for instance, is an event that brings together objects like buildings, voting machines, and TV sets, with actions like debates, news reporting, and people going to the polls. If we look at how all these various things serve as feedback paths for controlling the various participants’ perceptions, we’re on our way to a PCT analysis of this complex phenomenon.

Choosing one kind of feedback path instead of another for controlling a perception can be consequential in lots of ways, and I imagine you could come up with some examples of that yourself . . .

Best,

Kent

Kent

On Nov 2, 2012, at 6:14 PM, Richard Marken wrote:

[From Rick Marken (2012.11.02.1615)]

Kent McClelland (2012.11.02.1250)]

KM: As a sociologist I’m interested in looking at the environment as consisting of possible feedback paths for controlling different kinds of perceptions. I’ve done some thinking about how one might classify feedback paths, and I’d be happy to hear any corrections, additions, or suggestions:

RM: It would help me if you gave some concrete examples of each of these types of feedback paths. For example, I’d like a real world example of an exclusive feedback path and a non-exclusive one so I can see the difference. Right now I wouldn’t know on what basis to conclude that a person is controlling something via an exclusive feedback path, a flexible one, a high bandwidth one, and so on. Once I understand this (if I can;-) I might be able to provide some suggestions.

Best

Rick

Feedback paths provide the foundation for a sociological analysis from the PCT perspective, because social interaction takes place in a shared physical environment. Feedback paths can be analyzed on a number of dimensions.

    1.      Exclusivity—Inclusivity. Is this feedback path the “private property” of one individual or is it open to use by multiple individuals? How big is the pool of potential users? How many individuals depend on it for controlling their perceptions?
    2.      Flexibility of Purpose. How many different kinds of perceptions does this feedback path support? Is it single-purpose or multi-purpose? How many orders of perception can be supported by this feedback path? How many different perceptions at any given level does this feedback path support for the individual user?
    3.      Bandwidth. How strong is this feedback channel for maintaining control of a given perception? Feedback bandwidth may be increased by multiplying the effects of physical actions that compensate for disturbances, magnifying the quantity or quality of information passed to perceptual channels, or protecting the channel from possible disturbances.
    4.      Fixity—Evanescence. How nearly permanent or resistant to random physical forces is this feedback channel? Does it need to be reconstructed every time a given perception is controlled?
    5.      Manipulability. How easily can this feedback channel be manipulated (changed in a controlled way) to meet the objectives of the user?

Hypothesis: An individual’s liberty and flexibility of action is increased when the individual has access to feedback paths that are exclusive, multi-purpose, high bandwidth, and fixed but highly manipulable. (Which must be why people fall in love with the hand-held electronic devices that satisfy most of those requirements.)

Kent

Richard S. Marken PhD

rsmarken@gmail.com

www.mindreadings.com


Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[Martin Taylor 2012.11.04.14.08.CET]

It seems to me that Rick is insisting on limiting mathematics to

Arithmetic, whereas Kent wants to do Algebra or calculus. Rick wants
to know the specific values of things such as “what is the
controlled variable (variables?)” whereas Kent says “let there be a
controlled variable set x, y, z.” Rick says “How exactly does this
output affect the controlled variable at its input” whereas Kent
asks “what happens if pathways p, q,and/or r can can be used for the
outputs of control systems x, y, z to influence their perceptions”.
Kent’s approach looks rather more powerful when one is seeking
general statements such as “the locations with x,y values conforming
to x^2+y^2=c fall on a circle”. Rick’s is good when one is counting
change at the checkout and wants to know whether one should receive
$1.29 or $4.52. Rick’s method might eventually lead to an intuition
about the circle if he plotted enough events in which the sum of
squares of his particular x and y added up to c, but why would he
ever choose to look for such an odd thing? And how would he ever
show that the formula always works?
The difference between Kent’s approach and Rick’s seem analogous to
the difference between making generalizations from lots of random
experimental results in psychology and predicting what should be
expected using the principles of PCT.
Martin

···

[From Rick Marken (2012.11.04.0840)]

      Kent

McClelland (2012.11.03.1445)

      Rick Marken (2012.11.02.1615)


        > RM: It would help me if you gave some concrete examples

of each of these types of feedback paths. …

KM: Sure, I could provide some examples.

      1. Exclusive vs. Inclusive feedback paths: your private,

fenced-in backyard garden vs. a national park; a single-seater
ultralight plane vs. a jumbo jet that carries hundreds of
people.

      RM: Thanks, Kent. this is a good start. But my understanding

is that a feedback path is a connection between system outputs
and a controlled variable. So in order to get a better
understanding of your feedback paths I have to know what
variables you think are being controlled and the means people
use to control them…

      KM: From the

point of view of PCT, it seems to me, objects, actions and
events can all be analyzed in terms of feedback functions.

      RM: Again, not before you find out what variables are being

controlled and the outputs used to control them. Once you know
those things then the nature of feedback functions would,
indeed, contribute to our understanding of observable aspects
of human behavior.

[From Rick Marken (2012.11.04.1300)]

Martin Taylor( 2012.11.04.14.08.CET)–

      KM: From the

point of view of PCT, it seems to me, objects, actions and
events can all be analyzed in terms of feedback functions.

      RM: Again, not before you find out what variables are being

controlled and the outputs used to control them. Once you know
those things then the nature of feedback functions would,
indeed, contribute to our understanding of observable aspects
of human behavior.

MT: It seems to me that Rick is insisting on limiting mathematics to

Arithmetic, whereas Kent wants to do Algebra or calculus.

RM: Well, I was never all that good at math;-)

MT; Rick wants

to know the specific values of things such as “what is the
controlled variable (variables?)” whereas Kent says “let there be a
controlled variable set x, y, z.”

      RM: I missed where Kent said that. If Kent had actually givien values for x, y and z then I would have had a better understanding of why, say, a single-seater

ultralight plane is an exclusive feedback path while a jumbo jet is not. For example, if you are just controlling for getting from point A to point B then getting on either plane will produce this result; the number of other people who can ride with you (which may be what Kent meant by exclusivity) is irrelevant in this case. If, however, you are controlling for prestige, then which plane you get on (which feedback function you use) might matter.

I realize that my constant emphasis on the centrality of controlled variables to a PCT understanding of behavior conflicts with some pre-PCT concepts that you seem to want to cling to (like affordances). But you’ll just have to get over it;-) I’m sure you can do it.

Best

Rick

···
Rick says "How exactly does this

output affect the controlled variable at its input" whereas Kent
asks “what happens if pathways p, q,and/or r can can be used for the
outputs of control systems x, y, z to influence their perceptions”.

Kent's approach looks rather more powerful when one is seeking

general statements such as “the locations with x,y values conforming
to x^2+y^2=c fall on a circle”. Rick’s is good when one is counting
change at the checkout and wants to know whether one should receive
$1.29 or $4.52. Rick’s method might eventually lead to an intuition
about the circle if he plotted enough events in which the sum of
squares of his particular x and y added up to c, but why would he
ever choose to look for such an odd thing? And how would he ever
show that the formula always works?

The difference between Kent's approach and Rick's seem analogous to

the difference between making generalizations from lots of random
experimental results in psychology and predicting what should be
expected using the principles of PCT.

Martin


Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[Martin Taylor 2012.11.04.23.13 CET]

[From Rick Marken (2012.11.04.1300)]

        Martin Taylor(

2012.11.04.14.08.CET)–

                KM:

From the point of view of PCT, it seems to me,
objects, actions and events can all be analyzed in
terms of feedback functions.

                RM: Again, not before you find out what variables

are being controlled and the outputs used to control
them. Once you know those things then the nature of
feedback functions would, indeed, contribute to our
understanding of observable aspects of human
behavior.

        MT: It seems to me that Rick is insisting on limiting

mathematics to Arithmetic, whereas Kent wants to do Algebra
or calculus.

      RM: Well, I was never all that good at math;-)
We recognize that problem, and commiserate.
        MT; Rick wants to know

the specific values of things such as “what is the
controlled variable (variables?)” whereas Kent says “let
there be a controlled variable set x, y, z.”

      RM: I missed where Kent said that. If Kent had actually givien

values for x, y and z then I would have…

...been asking him to do simple arithmetic instead of the more

powerful algebra that could allow him to draw general conclusions
valid over a wide range of situations.

      had a better understanding of why, say, a single-seater

ultralight plane is an exclusive feedback path while a jumbo
jet is not.

The pilot controls for where the ultralight goes. The passenger in a

jumbo does not. Anyone who is controlling for going where the jumbo
happens to be going can use it as a means for controlling that
perception, but only if they are not controlling for going there at
some time other than when the jumbo flies. The pilot of the
ultralight has exclusive control of when and where the ultralight
goes (in the absence of overwhelming disturbances). The passengers
in the jumbo do not, and cannot influence it (Kent’s flexibility
characteristic) without resorting to the extreme of hijacking.

      I realize that my constant emphasis on the centrality of

controlled variables to a PCT understanding of behavior
conflicts with some pre-PCT concepts that you seem to want to
cling to (like affordances).

Whether that is true or false (and I think it is false) is

irrelevant to the issue I was addressing, which was why you and Kent
seem to be having trouble communicating. If you try to get over the
idea that there are no general principles in PCT, you might be able
to make a contribution his problem area.

Incidentally, can you show me where, outside of PCT, the concept of

environmental affordances has been used to mean the set of available
means by which an output can influence its controlled variable? Or
even where the word-pair has been used in any context outside of
PCT?

Martin

Hi,

I thought you changed Rick, as in one moment I thought you became self-critical

RM :

Actually, now that I think of it, I think I get into arguments (conflicts) so often on CSGNet because I am typically controlling for some perceptions (like PCT and liberalism) with such high gain that I tend to push back against disturbances to these perceptions rather forcefully (which just magnifies the conflict). But I can purposefully lower the gain when I want….

HB :

…but obviously I was wrong.

RM :

I realize that my constant emphasis on the centrality of controlled variables to a PCT understanding of behavior conflicts with some pre-PCT concepts that you seem to want to cling to (like affordances). But you’ll just have to get over it;-) I’m sure you can do it.

HB :

I think that “constant emphasis on the centrality to a PCT understanding” are not “controlled variables”, but their perception, as Martin emphasis : “output of control systems influence their perception”.

Maybe PCT is not about controlling “controlled variables” but “behavior is control of perception”. Whether variables are controlled in external environment or not (as I think that’s how you see it) is a big question” for about bilion reasons”.

It seems to me, that you want to control “controlled variable” with output ?

“Controlling variables” with output sounds to me so “behavioristic”. You will never know how you affected “controlled variable” if there is no sensory input, no control of perception.

Where in the PCT diagram (generic from Bill) in any of his books, you see “controlled variable” in it ? Maybe I missed something.

RM :

But my understanding is that a feedback path is a connection between system outputs and a controlled variable.

HB :

Maybe you understanding is wrong…

Bill P :

“That’s what feedback means : it’s an effect of system’s output on it’s own input” (LCS III, p.32, 2008)

HB :

Be usefull, Rick. There so much work to be done to improve PCT and you are “attacking” people who deserve the highest PCT respect. And I think also Bill respect them. So I don’t’ see the reason why you shouldn’t.

MT :

Rick says “How exactly does this output affect the controlled variable at its input”…

HB :

Did I miss something Martin ? I can’t find where Rick said that J

Best,

Boris

···

----- Original Message -----

From:
Richard Marken

To: CSGNET@LISTSERV.ILLINOIS.EDU

Sent: Sunday, November 04, 2012 10:00 PM

Subject: Re: Analyzing feedback paths

[From Rick Marken (2012.11.04.1300)]

Martin Taylor( 2012.11.04.14.08.CET)--
    KM: From the point of view of PCT, it seems to me, objects, actions and events can all be analyzed in terms of feedback functions.
  RM: Again, not before you find out what variables are being controlled and the outputs used to control them. Once you know those things then the nature of feedback functions would, indeed, contribute to our understanding of observable aspects of human behavior.
MT: It seems to me that Rick is insisting on limiting mathematics to Arithmetic, whereas Kent wants to do Algebra or calculus.

RM: Well, I was never all that good at math;-)

MT; Rick wants to know the specific values of things such as "what is the controlled variable (variables?)" whereas Kent says "let there be a controlled variable set x, y, z."

RM: I missed where Kent said that. If Kent had actually givien values for x, y and z then I would have had a better understanding of why, say, a single-seater ultralight plane is an exclusive feedback path while a jumbo jet is not. For example, if you are just controlling for getting from point A to point B then getting on either plane will produce this result; the number of other people who can ride with you (which may be what Kent meant by exclusivity) is irrelevant in this case. If, however, you are controlling for prestige, then which plane you get on (which feedback function you use) might matter.

I realize that my constant emphasis on the centrality of controlled variables to a PCT understanding of behavior conflicts with some pre-PCT concepts that you seem to want to cling to (like affordances). But you’ll just have to get over it;-) I’m sure you can do it.

Best

Rick

Rick says "How exactly does this output affect the controlled variable at its input" whereas Kent asks "what happens if pathways p, q,and/or r can can be used for the outputs of control systems x, y, z to influence their perceptions".

Kent's approach looks rather more powerful when one is seeking general statements such as "the locations with x,y values conforming to x^2+y^2=c fall on a circle". Rick's is good when one is counting change at the checkout and wants to know whether one should receive $1.29 or $4.52. Rick's method might eventually lead to an intuition about the circle if he plotted enough events in which the sum of squares of his particular x and y added up to c, but why would he ever choose to look for such an odd thing? And how would he ever show that the formula always works?

The difference between Kent's approach and Rick's seem analogous to the difference between making generalizations from lots of random experimental results in psychology and predicting what should be expected using the principles of PCT.

Martin


Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[From Kent McClelland (2012.11.04.17:15)

Relax, everyone. I’m not feeling at all attacked. I posted some half-baked ideas because they seemed interesting to me, and I was looking for criticism and corrections. When I get time tomorrow, I’ll try to respond to the interesting comments that Rick
and Martin have added to the thread. (I know that if I can get one of my ideas past Rick, it’s met a pretty rigorous standard. No shrinking violet he!)

(Sorry for the English slang, Boris. Shrinking violet just means someone who is timid and afraid to say anything. Obviously not Rick!)

Until tomorrow . . .

Kent

···

On Nov 4, 2012, at 5:03 PM, boris_upc wrote:

Hi,

I thought you changed Rick, as in one moment I thought you became self-critical…

RM :

Actually, now that I think of it, I think I get into arguments (conflicts) so often on CSGNet because I am typically controlling for some perceptions (like PCT and liberalism) with such high gain that I tend to push back against disturbances to these perceptions
rather forcefully (which just magnifies the conflict). But I can purposefully lower the gain when I want….

HB :

…but obviously I was wrong.

RM :

I realize that my constant emphasis on the centrality of controlled variables to a PCT understanding of behavior conflicts with some pre-PCT concepts that you seem to want to cling to (like affordances). But you’ll just have to get over it;-) I’m sure you can
do it.

HB :

I think that “constant emphasis on the centrality to a PCT understanding” are not “controlled variables”, but their perception, as Martin emphasis : “output of control systems influence their perception”.

Maybe PCT is not about controlling “controlled variables” but “behavior is control of perception”. Whether variables are controlled in external environment or not (as I think that’s how you see it) is a big question” for about bilion reasons”.

It seems to me, that you want to control “controlled variable” with output ?

“Controlling variables” with output sounds to me so “behavioristic”. You will never know how you affected “controlled variable” if there is no sensory input, no control of perception.

Where in the PCT diagram (generic from Bill) in any of his books, you see “controlled variable” in it ? Maybe I missed something.

RM :

But my understanding is that a feedback path is a connection between system outputs and a controlled variable.

HB :

Maybe you understanding is wrong…

Bill P :

“That’s what feedback means : it’s an effect of system’s output on it’s own input” (LCS III, p.32, 2008)

HB :

Be usefull, Rick. There so much work to be done to improve PCT and you are “attacking” people who deserve the highest PCT respect. And I think also Bill respect them. So I don’t’ see the reason why you shouldn’t.

MT :

Rick says “How exactly does this output affect the controlled variable at its input”…

HB :

Did I miss something Martin ? I can’t find where Rick said that J

Best,

Boris

----- Original Message -----

From: Richard Marken

To: CSGNET@LISTSERV.ILLINOIS.EDU

Sent: Sunday, November 04, 2012 10:00 PM

Subject: Re: Analyzing feedback paths

[From Rick Marken (2012.11.04.1300)]

Martin Taylor( 2012.11.04.14.08.CET)–

KM: From the point of view of PCT, it seems to me, objects, actions and events can all be analyzed in terms of feedback functions.

RM: Again, not before you find out what variables are being controlled and the outputs used to control them. Once you know those things then the nature of feedback functions would, indeed, contribute to our understanding of observable aspects of human behavior.

MT: It seems to me that Rick is insisting on limiting mathematics to Arithmetic, whereas Kent wants to do Algebra or calculus.

RM: Well, I was never all that good at math;-)

MT; Rick wants to know the specific values of things such as “what is the controlled variable (variables?)” whereas Kent says “let there be a controlled variable set x, y, z.”

RM: I missed where Kent said that. If Kent had actually givien values for x, y and z then I would have had a better understanding of why, say, a single-seater ultralight plane is an exclusive feedback path while a jumbo jet is not. For example, if you are just
controlling for getting from point A to point B then getting on either plane will produce this result; the number of other people who can ride with you (which may be what Kent meant by exclusivity) is irrelevant in this case. If, however, you are controlling
for prestige, then which plane you get on (which feedback function you use) might matter.

I realize that my constant emphasis on the centrality of controlled variables to a PCT understanding of behavior conflicts with some pre-PCT concepts that you seem to want to cling to (like affordances). But you’ll just have to get over it;-) I’m sure you can
do it.

Best

Rick

Rick says “How exactly does this output affect the controlled variable at its input” whereas Kent asks “what happens if pathways p, q,and/or r can can be used for the outputs of control systems x, y, z to influence their
perceptions”.

Kent’s approach looks rather more powerful when one is seeking general statements such as “the locations with x,y values conforming to x^2+y^2=c fall on a circle”. Rick’s is good when one is counting change at the checkout and wants to know whether one should
receive $1.29 or $4.52. Rick’s method might eventually lead to an intuition about the circle if he plotted enough events in which the sum of squares of his particular x and y added up to c, but why would he ever choose to look for such an odd thing? And how
would he ever show that the formula always works?

The difference between Kent’s approach and Rick’s seem analogous to the difference between making generalizations from lots of random experimental results in psychology and predicting what should be expected using the principles of PCT.

Martin

Richard S. Marken PhD

rsmarken@gmail.com

www.mindreadings.com

[From Rick Marken (2012.11.14.1530)]

Martin Taylor (2012.11.04.23.13 CET)–

      RM I missed where Kent said that. If Kent had actually givien

values for x, y and z then I would have had a better understanding of why, say, a single-seater
ultralight plane is an exclusive feedback path while a jumbo
jet is not.

MT: The pilot controls for where the ultralight goes. The passenger in a

jumbo does not. Anyone who is controlling for going where the jumbo
happens to be going can use it as a means for controlling that
perception, but only if they are not controlling for going there at
some time other than when the jumbo flies. The pilot of the
ultralight has exclusive control of when and where the ultralight
goes (in the absence of overwhelming disturbances). The passengers
in the jumbo do not, and cannot influence it (Kent’s flexibility
characteristic) without resorting to the extreme of hijacking.

RM: OK, so knowing that the controlled variable is getting from A to B we see that the exclusivity in the feedback function is the plane’s schedule, not its size (jumbo vs small private jet).

MT: Incidentally, can you show me where, outside of PCT, the concept of

environmental affordances has been used to mean the set of available
means by which an output can influence its controlled variable? Or
even where the word-pair has been used in any context outside of
PCT?

RM: Didn’t Gibson or one of his acolytes talk about a chair affording sitting? This is equivalent to saying that the chair can be the means to control a variable (body position). if you want to sit, the chair, with it’s horizontal seat, can be the means of achieving the goal of being in a seated body position (controlling for this variable). Of course, if your goal is something else, like breaking down a door, a chair can be the means of achieving that, too. Just a different aspect of the chair.

Best

Rick

···


Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[Martin Taylor 2012.11.05.06.55 CET]

[From Rick Marken (2012.11.14.1530)]

        Martin Taylor

(2012.11.04.23.13 CET)–

              RM I missed where Kent said that. If Kent had

actually givien values for x, y and z then I would
have had a better understanding of why, say, a
single-seater ultralight plane is an exclusive
feedback path while a jumbo jet is not.

        MT: The pilot controls for where the ultralight goes. The

passenger in a jumbo does not. Anyone who is controlling for
going where the jumbo happens to be going can use it as a
means for controlling that perception, but only if they are
not controlling for going there at some time other than when
the jumbo flies. The pilot of the ultralight has exclusive
control of when and where the ultralight goes (in the
absence of overwhelming disturbances). The passengers in the
jumbo do not, and cannot influence it (Kent’s flexibility
characteristic) without resorting to the extreme of
hijacking.

      RM: OK, so knowing that the controlled variable is getting

from A to B we see that the exclusivity in the feedback
function is the plane’s schedule, not its size (jumbo vs small
private jet).

The jumbo versus the ultralight isn't as good an example of

Exclusivity as Kent’s other example of one’s backyard versus a
National Park, because it also differs too much on his Flexibility
dimension. I used the flexibility characteristic improperly in my
reply.

I prefer the private driveway versus the public road as an example

of exclusivity. You are the only one allowed to determine who parks
in your driveway, but you and everyone else can equally use the
roadway. That is true also of your ultralight versus the jumbo.
(Incidentally, another aspect of the ultralight’s flexibility is
that it doesn’t have to go from A to B. It can also go from A to C
if you want it to. You can’t do that with the jumbo.)

If I understand Kent's point about exclusivity, you have more

liberty if you know you have exclusive access to an environmental
affordance than if someone else may use it at the same time. Using
your example from the end of this message, if someone is using a
chair to break down a door, you can’t sit on it. To me, the
underlying issue is the exclusive use of degrees of freedom
(technical use of the word “freedom” here) in the path. If someone
else is using some components of a feedback path and leaves fewer
available degrees of freedom for you, you can’t use that path to
control that variable as effectively as you might otherwise have
done.

Aside...

[From Bill Powers (2012.11.04.1255 MST)] Gibson's concept of

“affordances” (“the opportunities for
action provided by a particular object or environment”) is
exceedingly naive, “Opportunities” is a qualitative notion,
with the number of different opportunities provided by any single
affordance, when expressed quantitatively, being for all practical
purposes infinite.

[MT now] I was once on a platform with Gibson, long ago, and said

much the same thing to his face, without using the personal
comments.

[BP] Furthermore, since organisms control not actions but

consequences of actions, an affordance can actually relate to any
number
of different controlled variables which are affected via similar
feedback
functions.

[MT now] I would call that "naive", since you like the word. A round

stone is not easy to use to contain liquid or to clean your teeth.
It is not an environmental affordance if you are controlling for
seeing liquid kept safely or for having clean teeth. A stone is, but
a glass jug usually is not an environmental affordance if you want
to hammer in a nail. Kent doesn’t throw up his hands and say
“anything can be used for any purpose, so let’s not consider the
question”. His dimension of Flexibility tries to deal with such
limitations, arguing that wider limits on what you can use something
to control leads to more freedom of action.

End Aside...

Back to Kent's suggested factors that affect "liberty and

flexibility of action". Rick questioned only the
“Exclusivity-Inclusivity” factor. The next was “Flexibility of
Purpose”, which Bill says is “for all practical
purposes infinite”. To my mind, it is precisely for practical
purposes that the possibilities for using particular features of the
environment are not infinite. One type of puzzle is to provide a set
of objects that a clever creative person could put together to
control some perception. In the view of the puzzle-setter, those
objects may together provide an environmental affordance for that
particular control system, but it is not for the puzzle solver until
the puzzle is solved, nor is it likely to be a practical one unless
the person absolutely must control that variable and can find no
other environmental affordance to do it. But to follow Bill, almost
anything in the person’s environment could be such an affordance. So
why is there a puzzle?

Kent's third was something he called "Bandwidth". I would prefer to

call it “Power, Precision and Speed” (PoPS), all of which relate to
how effectively the output can counter disturbances. A faster path
permits control against faster changing disturbances, a more precise
path permits more exact control over even slowly changing
disturbances, and a more powerful path permits control over a wider
range of disturbance values. Increases in any of these values seems
to me to increase a person’s liberty (though possibly not the
liberty of others in the person’s social environment).

Kent's fourth dimension was "Fixity-Evanescence". Is the bridge

guaranteed to be open for traffic when you arrive, or do you have to
wake the bridgemaster to swing it your way because he may have it
set for river traffic? Control is easier if you can rely on the
output having an effect on the controlled variable, rather than
having to change mechanisms or reorganize when the pathway is broken
and your output does not affect your perceptual variable.

The fifth dimension was "Manipulability". I think this ties in with

the fourth. Both require perceptual control of control loops. To
manipulate a feedback pathway so as to provide an environmental
affordance for control of some other perception requires a
perception of that other control loop and a reference value that
corresponds to “the other control loop functioning well”. It is
clearly something we do, but not within the classical hierarchy of
control, so far as I can see. However, it is also not clear to me
that “Manipulability” of a feedback path for a single control unit
is really what Kent is (or should be) dealing with, although grading
a street to remove potholes does qualify. I think the issue is more
the manipulability of the environment in the manner of railway
switches that alter the environment so that at some moments one
perception can be controlled and at other moments others can be
controlled. I’m not at all clear on this one, and maybe Kent will
clarify it.

Kent's "Hypothesis: An individual’s liberty and flexibility of

action is increased when the individual has access to feedback paths
that are exclusive, multi-purpose, high bandwidth, and fixed but
highly manipulable" seems quite reasonable to me, but from the
“Social PCT” point of view the effects on other people en masse and
as interacting individuals need analysis. There is, after all, a
reason why jumbo jets are built and why people use them. I equate
“liberty and flexibility of action” with the ability to control a
wide range of perceptions in the presence of fast and powerful
disturbances. I don’t know if that is how Kent thinks of it.

          MT: Incidentally, can you show me where,

outside of PCT, the concept of environmental affordances
has been used to mean the set of available means by which
an output can influence its controlled variable? Or even
where the word-pair has been used in any context outside
of PCT?

  RM: Didn't Gibson or one of his acolytes talk about a chair

affording sitting? This is equivalent to saying that the chair can
be the means to control a variable (body position). if you want
to sit, the chair, with it’s horizontal seat, can be the means of
achieving the goal of being in a seated body position (controlling
for this variable). Of course, if your goal is something else,
like breaking down a door, a chair can be the means of achieving
that, too. Just a different aspect of the chair.

Did he? I don't remember that. But only someone thinking in terms of

PCT would use the equivalence you did, and using that equivalence he
would have been correct. Sitting is not possible in the absence of a
reasonably horizontal surface. You don’t have to know PCT to know
that. But you do have to know PCT to recognize that the horizontal
surface is a necessary part of a control loop controlling the
perception of a particular body posture. And you would have to be
thinking PCT to use the term “environmental affordance” rather than
“affordance”.

Martin

[From Bill Powers (2012.11.05.0702 MST)]

Martin Taylor 2012.11.05.06.55
CET

[BP earlier] Furthermore, since
organisms control not actions but consequences of actions, an affordance
can actually relate to any number of different controlled variables which
are affected via similar feedback functions.

[MT] I would call that “naive”, since you like the word. A
round stone is not easy to use to contain liquid or to clean your teeth.
It is not an environmental affordance if you are controlling for seeing
liquid kept safely or for having clean teeth. A stone is, but a glass jug
usually is not an environmental affordance if you want to hammer in a
nail. Kent doesn’t throw up his hands and say “anything can be used
for any purpose, so let’s not consider the question”. His dimension
of Flexibility tries to deal with such limitations, arguing that wider
limits on what you can use something to control leads to more freedom of
action.

BP: My differentiation between qualitative and quantitative affordance
was meant to point out that an affordance is usually spoken of not as a
specific property of a feedback path, but as a kind or class of
properties. I have in fact used a round stone to see that " a liquid
is kept safely" (as you generalized the idea), using it to weigh
down the lid over a glass of something like milk to keep the wind from
blowing it off at a picnic, letting the dust in.
A feedback path, as Rick keeps pointing out, is modeled as connecting an
output quantity to a region in the environment in which a control system
perceives a controlled variable. The controlled variable, if discoverable
by others, is some function of the individual physical variables that
others can see in that part of the environment. The nature of that
function is determined by the organization of the controlling system’s
input function, not the organization of the environment or the feedback
path. I can decide to control the sum (or product or quotient) of the
Fahrenheight temperature reading on a thermometer and the reading of the
minute hand of an alarm clock, by acting on the thermometer or the alarm
clock or both. Could that affordance be predicted by anyone? Is it a
self-evident property of the physical world? If I notice that I can also
control the distance between the thermometer and the alarm clock by
acting on either or both, does this mean I have somehow changed the
affordances that exist in the environment?
So an affordance is literally any way of controlling any kind of variable
that one can imagine. This suggests that the number of unused affordances
is infinite.
What is “naive” about the concept of affordance is the realism
assumed. One simply has to refrain from noticing that the items which
appear to have these affordances are also perceptions, existing not
actually in the environment but inside the controlling or observing
system. In order to conclude that perception in general is
“direct” as Gibson does, it is necessary first to accept the
premise that it is direct, for all of Gibson’s attempts to show that one
perception is direct depend on accepting that other perceptions are
direct – the “flatness” of a “plane surface” for
example. So Gibson’s primary logical error is petitio principii,
the error called “begging the question,” or “affirming the
consequent.” One can say that a perception is “in the
light” only if one fails to notice that “light” is also a
perception ( as is
“in”).

[S]eldom is anyone going to simply place the conclusion
word-for-word into the premises … Rather, an arguer might use phraseology
that conceals the fact that the conclusion is masquerading as a premise.
The conclusion is rephrased to look different and is then placed in the
premises.
­Paul Herrick

http://en.wikipedia.org/wiki/Begging_the_question

I think that is a fair description of what Gibson did.

Best,

Bill P.

[From Bill Powers (2012.11.05.1752 MST)]

Kent McClelland (2012.11.02.1250) --

As a sociologist I'm interested in looking at the environment as consisting of possible feedback paths for controlling different kinds of perceptions. I've done some thinking about how one might classify feedback paths, and I'd be happy to hear any corrections, additions, or suggestions:

Feedback paths provide the foundation for a sociological analysis from the PCT perspective, because social interaction takes place in a shared physical environment. Feedback paths can be analyzed on a number of dimensions:

BP: Maybe now I can make my proposal clearer. Rick Marken is quite right in saying that before you can define a feedback path, you have to identify the variable under control, because the same output can be connected by the same physical path to any number of different controlled variables that the controlling system sees in the same part of the environment. There is no such thing as a variable that is inherently "salient." If there are two physical variables, x and y, in the envronment, how many possible controlled variables are there? The number is the same as the number of independent functions of two variables that exists, minus the ones impossible to control. And to control each one requires adjusting the weightings of the output quantities, as well as the form of the output function, as required to achieve negative feedback.

Before you discuss the effect of various feedback paths on the ability to control, I propose, it is necessary first to say why control of the particular variable involved is needed. Why do people control anything at all? This question can't be answered by describing feedback paths. To the family who clean my house every two weeks, the difference in flexibility afforded by jumbo jets versus private aircraft is completely irrelevant, since they don't have enough money to travel either way (I almost said "they can't afford to...".)

Why does a person control anything at all? I think we can at least make a list of necessities as opposed to discretionary control processes. A person must eat, stay warm, stay safe, stay healthy, breathe, and so on. These requirements are not defined by the feedback paths, but by a hierarchy of goals ranging from gene expressions to system concepts. You can make a list of affordances that would be sufficient in a given environment to allow the necessary control to be carried out, but only in general terms. In our society, the person must have some way to acquire food, for example, and that is most commonly by having a job that earns a wage. When you think in those terms, "unemployment" takes on far more meaning than it has in the monthly jobs report. As one character in a political ad said, approximately, "They told us the plant was closed and we were all fired. We had no income any more. It made me sick."

Having no income means that you can't buy food, clothing, warmth, shelter, and so on down the list of necessities. The average person has savings enough to last for two weeks without income. If that is all that happens, you can only prepare to die, or try other things to get those items, such as crime. It's really quite simple. The economists say, "Well, they can just move to a state where there are some jobs available," which in French comes out as "Let them eat cake." Talk about naive! I remember my mother in the 1930s making a pile of sandwiches every few days, then taking them to the back door to give to the nomadic hoboes who came by several times a day, seldom the same ones. Let them eat sandwiches -- if someone can afford to give them away, which my family couldn't afford, but did anyway.

I think this is the level where we have to start to get a theory that is as close as possible to real life. That's how you get to a theory that works for everyone every time.

Best,

Bill P.

[From Kent McClelland (2012.11.05.0935)]

Thank you to Martin, Rick, and Boris for comments on the thread that I started about analyzing feedback paths. Martin has been arguing, probably more ably than I can do, on behalf of the position I tried to lay out, but I have some responses to some of the points made in various posts.

Martin Taylor (2012.11.03.08.13 CET)

MT: Seeing your post reminded me that before I left home I was working on an answer to your "more about gain" posting in which you asked about the "composite feedback loop". The algebra isn't complicated, but there are a lot of symbols and I didn't trust my answers so I didn't post it before I went to Europe. I intended to post it just now after seeing your name, including errors that probably persist, but I find that the diagrams are held on my desktop machine, not this one, so I can't post until I get home in a couple of weeks.

KM: Thank you, Martin. I will look forward to seeing your post. I kind of dropped that earlier thread myself when I got working on other things, but it's definitely something I'd like to get back to.

MT: All of your points are excellent, except that they refer to "individuals", by which I assume individual animals. They seem mostly to apply to individual control units, though the final hypothesis deals with the collection of control units in an individual animal.

KM: Good point. Yes, I definitely meant the analysis to apply to other animals, as well as humans.

MT: [Exclusivity�Inclusivity dimension.] A feedback path that can be used only by one control unit is not susceptible to disturbance from the directly conflicting actions of others. It is, however, subject to disturbance from the side-effects of the actions of others (the situation I am addressing in my response to your "more about gain" posting. So it is possible that even if the feedback path is used only by the one control unit, its disturbance is influenced by all the other control units in the world. To get to the "private property" equivalent, you have to shield the feedback path. At the level of individuals, we have laws that effect the shielding. If I own something, and I say you can't do anything to influence it, there may be legal reasons why you shouldn't, so you probably won't. But laws just reduce the probability, and in any case it is usually hard for either of use to know whether the side effects of your legitimate activity will disturb my control unit.

KM: Again, good point. For this kind of analysis it's certainly important to consider the shielding of feedback paths, or in other words reduction of the amplitude and frequency of disturbances in those paths, particularly for feedback paths that are the "private property" of one individual.

MT: [Flexibility of Purpose.] You are talking about what I have called "environmental affordances". A stone could serve as a part of an ornamental display, a pounder for driving in a stake, a projectile for throwing, and so forth. However, when it is used for one of these purposes, it cannot be used for another. If only that stone is available and at some moment the individual is trying to control two of those perceptions, there is conflict. If there is a heavy hammer at hand, and more throwable stones, then this pretty one can continue to serve its function in your feedback loop for controlling your perception of the ornamental display, eliminating the conflict. So "Flexibility of Purpose" must be considered along with the ranges of means by which all these other purposes can be achieved. The more environmental affordances associated with a particular feedback path, the more opportunity for conflict (it's the direct opposite of your "privacy" state), and the more environmental affordances in the environment for a particular purpose, the more opportunity for avoiding conflict. We usually discuss this under the heading "degrees of freedom".

KM: All good points, except that I'd prefer to skip the "affordances" terminology, which just gets people confused with Gibson's perspective, and I agree with Bill that it's important to keep the PCT perspective separate from Gibson's approach. What I'd say about the flexibility dimension is that some "tools" are designed to be multipurpose, while others are designed with a single purpose in mind (and here I'm equating feedback paths through objects or artifacts with the idea of using a tool, which seems like a useful way to think about it). And although any object or artifact can potentially serve as a feedback path for controlling more than one perception, either in terms of several perceptions controlled at the same time by a single individual or several different perceptions controlled by multiple individuals using that object, my point here has to do with the design or intended use envisioned by the maker of the artifact.

MT: "Bandwidth" has an engineering meaning relating to how fast unpredictable changes can be propagated through a channel. That's the meaning appropriate to PCT discussions. You mention several different properties that may be important, such as the range and precision of possible effects of the output upon the environmental variable affected by a disturbance, the speed and precision of the perceptual pathway from the environment to the perceptual signal value, the shielding I mentioned in respect to your first point. All of these do affect how well the control system can control over a wide range of disturbance values, or, using a language closer to yours, how reliable is the feedback path.

KM: OK, bandwidth was a poor choice of terminology on my part, since I was using the term as a loose analogy and ignoring its engineering definition. Your substitution of the ideas of range, precision, and speed gets certainly closer to my intended meaning. Thinking more about this dimension, I still haven't settled on a single term to encapsulate the idea. Maybe "effectiveness" of the feedback path is what I'm driving at. Does the path get you a big, reliable dose of whatever you're looking for (the CEV)? Another way to talk about it might be to describe it as feedback "throughput." Does your output lead to strong input of the desired perception? Perhaps this idea is just a matter of the amount of gain in the feedback function? Speaking in practical terms, any tools that amplify your output (like power tools), or your input (like scientific instruments that allow observation of very small or very distant phenomena), or just carry your output across distances (like microwaves that carry your phone conversations), would be the kinds of effectiveness-enhanced feedback paths I'm thinking of.

MT: [Fixity�Evanescence.] Here you get into a pervasive issue in PCT: the interactions between control systems that act on their environment to control their perceptions (all control systems) and those whose perceptions are the working components of other control systems. A control system doesn't know anything about its feedback path. All that happens if the feedback path is broken is that control fails. In much PCT discussion, if control persistently fails, "reorganization" happens, and that is a random process. This doesn't seem to me to be what happens in real life. If I use a bridge on my regular route, and it is closed for maintenance one day, I may cast about until I find an alternate route. This alternate route works to control the perceptions relating to getting where I am going, so why should I go back to using the bridge when I learn that the maintenance work is complete? There must be some other perception whose control affects part of the "going to that place" set of control systems. How that works is something that needs to be addressed within PCT.

KM: Speaking from my own experience, I'd say that when control of a perception fails, the next thing that ordinarily happens is a rapid review in imagination of other ways to control that perception (or, in other words, other possible feedback paths, to use the terms of this discussion) and then the quick substitution of a line of action that is imagined to yield a positive outcome. It's only when control of the perception fails repeatedly, for all the outputs one can think of, so that higher-level perceptions like "I've got to go to work today" are threatened, that reorganization begins.

MT: [Manipulability.] This also deals with the question of how one control system can control perceptions relating to the performance of another. Who (which control system) does the manipulating, in control of what perception? It is not control system whose feedback channel is being manipulated. "The objectives of the user" form one or more controlled perceptions that in their output pathways incorporate lower-level control systems. Your question presumes that at least one of these lower level systems would work better if its feedback path were manipulated. Who does the manipulating and why?

KM: On this dimension, again, it seems that I haven't done a good job of selecting my terms. I'm not specifically talking about the manipulation of one person by another (though that might be something that one would want to consider when investigating a situation in which the feedback path for someone to get something done goes through another person or persons). Rather, I'm talking about something like the malleability of an object to be used as a feedback path, or the amenability of the object to being quickly modified to suit the present occasion. If it isn't quite the right tool for the job, can you change the "settings" to make it more suitable? A computer that you can program, assuming the use of the programming language isn't too arduous, is my idea of an object that would score high on this dimension (whatever we might call it). Smart phones with lots of apps are pretty good on this, too.

A side comment: although I've been talking strictly about objects and artifacts, human bodies, when viewed as physical objects, either one's own body or others', can serve as feedback paths that rank high on these dimensions: exclusivity (assuming you're not a slave), flexibility, effectiveness, fixity, and malleability--all told, a highly useful tool for controlling all kinds (literally) of perceptions. The problems of using other people's bodies as feedback paths for controlling your own perceptions are related, of course, to the issues of conflict and alignment of reference standards that we've talked about in other threads.

This post is getting pretty long, so I'll just send it out and go on to respond to other comments from the thread in a later post.

My best to all,

Kent

···

On Nov 3, 2012, at 3:22 AM, Martin Taylor wrote:

[Martin Taylor 2012.11.03.08.13 CET]

[From Kent McClelland (2012.11.02.1250)]

Seeing your post reminded me that before I left home I was working on an answer to your "more about gain" posting in which you asked about the "composite feedback loop". The algebra isn't complicated, but there are a lot of symbols and I didn't trust my answers so I didn't post it before I went to Europe. I intended to post it just now after seeing your name, including errors that probably persist, but I find that the diagrams are held on my desktop machine, not this one, so I can't post until I get home in a couple of weeks.

As a sociologist I'm interested in looking at the environment as consisting of possible feedback paths for controlling different kinds of perceptions. I've done some thinking about how one might classify feedback paths, and I'd be happy to hear any corrections, additions, or suggestions:

All of your points are excellent, except that they refer to "individuals", by which I assume individual animals. They seem mostly to apply to individual control units, though the final hypothesis deals with the collection of control units in an individual animal.

Feedback paths provide the foundation for a sociological analysis from the PCT perspective, because social interaction takes place in a shared physical environment. Feedback paths can be analyzed on a number of dimensions:

  1. Exclusivity�Inclusivity. Is this feedback path the �private property� of one individual or is it open to use by multiple individuals? How big is the pool of potential users? How many individuals depend on it for controlling their perceptions?

A feedback path that can be used only by one control unit is not susceptible to disturbance from the directly conflicting actions of others. It is, however, subject to disturbance from the side-effects of the actions of others (the situation I am addressing in my response to your "more about gain" posting. So it is possible that even if the feedback path is used only by the one control unit, its disturbance is influenced by all the other control units in the world. To get to the "private property" equivalent, you have to shield the feedback path. At the level of individuals, we have laws that effect the shielding. If I own something, and I say you can't do anything to influence it, there may be legal reasons why you shouldn't, so you probably won't. But laws just reduce the probability, and in any case it is usually hard for either of use to know whether the side effects of your legitimate activity will disturb my control unit.

  2. Flexibility of Purpose. How many different kinds of perceptions does this feedback path support? Is it single-purpose or multi-purpose? How many orders of perception can be supported by this feedback path? How many different perceptions at any given level does this feedback path support for the individual user?

You are talking about what I have called "environmental affordances". A stone could serve as a part of an ornamental display, a pounder for driving in a stake, a projectile for throwing, and so forth. However, when it is used for one of these purposes, it cannot be used for another. If only that stone is available and at some moment the individual is trying to control two of those perceptions, there is conflict. If there is a heavy hammer at hand, and more throwable stones, then this pretty one can continue to serve its function in your feedback loop for controlling your perception of the ornamental display, eliminating the conflict. So "Flexibility of Purpose" must be considered along with the ranges of means by which all these other purposes can be achieved. The more environmental affordances associated with a particular feedback path, the more opportunity for conflict (it's the direct opposite of your "privacy" state), and the more environmental affordances in the environment for a particular purpose, the more opportunity for avoiding conflict. We usually discuss this under the heading "degrees of freedom".

  3. Bandwidth. How strong is this feedback channel for maintaining control of a given perception? Feedback bandwidth may be increased by multiplying the effects of physical actions that compensate for disturbances, magnifying the quantity or quality of information passed to perceptual channels, or protecting the channel from possible disturbances.

"Bandwidth" has an engineering meaning relating to how fast unpredictable changes can be propagated through a channel. That's the meaning appropriate to PCT discussions. You mention several different properties that may be important, such as the range and precision of possible effects of the output upon the environmental variable affected by a disturbance, the speed and precision of the perceptual pathway from the environment to the perceptual signal value, the shielding I mentioned in respect to your first point. All of these do affect how well the control system can control over a wide range of disturbance values, or, using a language closer to yours, how reliable is the feedback path.

  4. Fixity�Evanescence. How nearly permanent or resistant to random physical forces is this feedback channel? Does it need to be reconstructed every time a given perception is controlled?

Here you get into a pervasive issue in PCT: the interactions between control systems that act on their environment to control their perceptions (all control systems) and those whose perceptions are the working components of other control systems. A control system doesn't know anything about its feedback path. All that happens if the feedback path is broken is that control fails. In much PCT discussion, if control persistently fails, "reorganization" happens, and that is a random process. This doesn't seem to me to be what happens in real life. If I use a bridge on my regular route, and it is closed for maintenance one day, I may cast about until I find an alternate route. This alternate route works to control the perceptions relating to getting where I am going, so why should I go back to using the bridge when I learn that the maintenance work is complete? There must be some other perception whose control affects part of the "going to that place" set of control systems. How that works is something that needs to be addressed within PCT.

  5. Manipulability. How easily can this feedback channel be manipulated (changed in a controlled way) to meet the objectives of the user?

This also deals with the question of how one control system can control perceptions relating to the performance of another. Who (which control system) does the manipulating, in control of what perception? It is not control system whose feedback channel is being manipulated. "The objectives of the user" form one or more controlled perceptions that in their output pathways incorporate lower-level control systems. Your question presumes that at least one of these lower level systems would work better if its feedback path were manipulated. Who does the manipulating and why?

Hypothesis: An individual�s liberty and flexibility of action is increased when the individual has access to feedback paths that are exclusive, multi-purpose, high bandwidth, and fixed but highly manipulable. (Which must be why people fall in love with the hand-held electronic devices that satisfy most of those requirements.)

Kent

Apart from the use of the word "bandwidth" to encompass speed, range, and precision, that hypothesis seems most reasonable.

The word "access" is important, as it implies not only that an outside observer could know that a particular feedback path is available for the individual to control a particular perception, but that the individual has reorganized to use it when appropriate (not conflicting with other purposes -- see the example of the stone, above.

"Access" very often means "money", often enough that the possibility of other means is ignored. Money can limit access for some purposes, just as it increases access for most. The old saw says "money can not buy happiness", but it can buy many possibilities for control. There used to be a cartoon series "Happiness is a ..." (e.g. warm hug). So what characteristic in PCT corresponds to "happiness"?

Martin

[From Kent McClelland (2012.11.05.1350)]

Here are some more of my thoughts about previous posts on this thread.

Rick Marken (2012.11.04.0840)

KM: In response to my examples of feedback paths that differ along the five dimensions I set out, Rick writes:

RM: Thanks, Kent. this is a good start. But my understanding is that a feedback path is a connection between system outputs and a controlled variable. So in order to get a better understanding of your feedback paths I have to know what variables you think
are being controlled and the means people use to control them. Once I know these things – controlled variables and the means available for controlling them – I think I’ll be able to understand better your classification of the feedback paths connecting
them.

And I responded. . .

KM: Backing up, the reason I’m focusing on feedback paths is because it seems like people interested in PCT have already done a pretty good job of exploring the psychological workings of the control loops inside a person’s body, looking at perceptual input
functions, comparator functions, output functions, and the various signals that connect these functions. But what makes PCT approach distinctive is that we insist that it’s also important to consider feedback functions. However, my sense of where we’re at
is that the PCT analysis of feedback functions is relatively undeveloped, other than paying close attention to the influence of disturbances.

RM: I think what makes PCT distinctive is the emphasis on the fact that behavior is organized around the control of perceptual variables (controlled variables) and that understanding behavior is really a matter of learning what these variables are and
how they are controlled. I think one has to do research aimed at the discovery of controlled variables before it makes any sense to analyze feedback functions since (as I said above) feedback functions are connections between system outputs and controlled
variables. If you don’t know what variables are being controlled you can’t possibly know which outputs are affecting them or how they are affecting them (the nature of the feedback function).

KM: In describing the properties of feedback paths, I have been talking mostly about manufactured objects and artifacts, and for many of these things their intended uses (perceptions that the user is seeking to control) are obvious to anyone who is culturally
competent. It doesn’t take an elaborate test for the controlled variable to understand the perceptions that the ordinary users of these artifacts are trying to control. And the feedback paths that I’m talking about (in other words, the artifacts in question)
are precisely the means that people are using to control these perceptions.

Let’s consider artifacts like a dishwasher and a baseball glove. Of course, a dishwashing machine or a baseball glove could potentially be used for a myriad of purposes depending on the creativity and special interests of the user, but 99 percent of the time
the dishwasher will be used to clean dirty dishes and the baseball glove to catch baseballs. If you saw someone putting frozen foods in a dishwasher and dirty dishes in a freezer, you might think that doing a test for the controlled variable was in order,
and likewise, if someone were wearing a baseball glove on his head as he tried to snag a baseball with his hat, but most manufactured objects have been designed for specific uses or a set of uses that are not too difficult to figure out (or at least not too
difficult until you read the instruction booklet that comes along with them, at which point you might be really confused!).

In your first paragraph above, you describe a feedback path as the connection between system outputs and the controlled variable in the environment, but I would have thought that a better name for that segment of a control loop would be the feedback function.
By feedback path, I’m referring to the whole segment of the control loop from the system’s output to the input, and in my view the controlled variable can be considered part of the feedback path.

This distinction between feedback function and feedback path can be an important one, particularly if the controlled variable of interest to the user of the feedback path has been brought into control largely or entirely by the efforts of other living individuals
or mechanical control systems, a condition we encounter quite frequently when we take the PCT analysis out of the psychological laboratory and attempt to apply it to social systems in the wild, as it were. Let me give an example.

Suppose you have a favorite television program, and you really like to experience the perception of yourself enjoying the kind of entertainment it provides. What can you do if you want to experience that perception again? Option A: Think back over episodes
of the program you have seen and construct an elaborate imaginary story in your head that stars the regulars on the show, then sit back and watch it unfold on your “eyelid movies.” Option B: Turn on the TV set when the program comes on (or look up an episode
on Hulu) and watch an episode that a big team of actors, directors, producers, camera operators, TV executives, makeup artists, accountants, sales execs, etc., etc., has prepared for your enjoyment. These are two different feedback paths for controlling the
same (or very similar) perceptions.

Which option requires more effort on the part of the user? Which option is likely to provide a more satisfactory experience? How much real controlling (and of what variables) does the individual in question need to do in order to experience the controlled variable
of interest–the experience of being entertained-- when it is handed to him or her as a prepackaged episode? I expect that 99 out of 100 individuals wanting to have that experience will choose option B, and I would regard the TV episode as one component of
a commonly used feedback path for satisfying that kind of desire for entertainment.

only understood PCT). Questions like (1) what are the consequences if lots of people choose one kind of feedback path instead of another to control a given kind of perception (as when most people drive to work instead of taking mass transit), and (2) when
we see people controlling obvious perceptions by using stereotypical feedback paths, what other perceptions may be “motivating” that behavior (or in other words, what may be the higher-level perceptions for which their obvious control of the lower-level perceptions
can be a means to an end).

Let me say again, if we want to spread the word about PCT among social scientists, we need to focus on a level of analysis that they are really interested in, and the predominant focus of PCT scholars on what happens inside the person’s head, while paying less
attention to what happens in the person’s environment, has made it harder for us to communicate with many social scientists.

Kent

···

from my point of view as a sociologist, there are more interesting questions than which perceptions are people controlling when they use familiar objects in the culturally expected ways (and I’m speaking, I think, on behalf of most social scientists, if they

On Nov 4, 2012, at 10:43 AM, Richard Marken wrote:

[From Rick Marken (2012.11.04.0840)]

Kent McClelland (2012.11.03.1445)

Rick Marken (2012.11.02.1615)

RM: It would help me if you gave some concrete examples of each of these types of feedback paths. …

KM: Sure, I could provide some examples.

  1. Exclusive vs. Inclusive feedback paths: your private, fenced-in backyard garden vs. a national park; a single-seater ultralight plane vs. a jumbo jet that carries hundreds of people.

RM: Thanks, Kent. this is a good start. But my understanding is that a feedback path is a connection between system outputs and a controlled variable. So in order to get a better understanding of your feedback paths I have to know what variables you think are
being controlled and the means people use to control them. Once I know these things – controlled variables and the means available for controlling them – I think I’ll be able to understand better your classification of the feedback paths connecting them.

KM: Backing up, the reason I’m focusing on feedback paths is because it seems like people interested in PCT have already done a pretty good job of exploring the psychological workings of the control loops inside a person’s body, looking at perceptual input
functions, comparator functions, output functions, and the various signals that connect these functions. But what makes PCT approach distinctive is that we insist that it’s also important to consider feedback functions. However, my sense of where we’re at
is that the PCT analysis of feedback functions is relatively undeveloped, other than paying close attention to the influence of disturbances.

RM: I think what makes PCT distinctive is the emphasis on the fact that behavior is organized around the control of perceptual variables (controlled variables) and that understanding behavior is really a matter of learning what these variables are and how they
are controlled. I think one has to do research aimed at the discovery of controlled variables before it makes any sense to analyze feedback functions since (as I said above) feedback functions are connections between system outputs and controlled variables.
If you don’t know what variables are being controlled you can’t possibly know which outputs are affecting them or how they are affecting them (the nature of the feedback function).

KM: Most social scientists–including most economists, political scientists, sociologists, and anthropologists–aren’t particularly interested in the psychological details of how neural connections work. Instead, they are intensely interested in understanding
and explaining empirically observable objects, patterns of human action, and events that occur outside the human body.

RM: I agree. And I think that what PCT shows is that in order to understand these things social scientists have to understand what variables people are controlling, how they control them (the lower order means used) and why (the higher order reasons for controlling
the variable at particular levels).

KM: From the point of view of PCT, it seems to me, objects, actions and events can all be analyzed in terms of feedback functions.

RM: Again, not before you find out what variables are being controlled and the outputs used to control them. Once you know those things then the nature of feedback functions would, indeed, contribute to our understanding of observable aspects of human behavior.

Best

Rick

In the context of control loops, objects, actions, and events serve to provide feedback paths for controlling perceptions. If we can focus on explaining objects, actions, and events from a PCT perspective, we’ll be meeting social scientists more than halfway.

The examples I’ve given above are all examples of objects used as feedback paths. In economic lingo, I’m talking about goods, and I think we could make a similar analysis of services: the patterns of action that serve as feedback paths for controlling the perceptions
people want to control. Events might be a little trickier to analyze, except that they often occur as a confluence of objects and actions. An election, for instance, is an event that brings together objects like buildings, voting machines, and TV sets, with
actions like debates, news reporting, and people going to the polls. If we look at how all these various things serve as feedback paths for controlling the various participants’ perceptions, we’re on our way to a PCT analysis of this complex phenomenon.

Choosing one kind of feedback path instead of another for controlling a perception can be consequential in lots of ways, and I imagine you could come up with some examples of that yourself . . .

Best,

Kent

Kent

On Nov 2, 2012, at 6:14 PM, Richard Marken wrote:

[From Rick Marken (2012.11.02.1615)]

Kent McClelland (2012.11.02.1250)]

KM: As a sociologist I’m interested in looking at the environment as consisting of possible feedback paths for controlling different kinds of perceptions. I’ve done some thinking about how one might classify feedback paths, and I’d be happy to hear any corrections,
additions, or suggestions:

RM: It would help me if you gave some concrete examples of each of these types of feedback paths. For example, I’d like a real world example of an exclusive feedback path and a non-exclusive one so I can see the difference. Right now I wouldn’t know on what
basis to conclude that a person is controlling something via an exclusive feedback path, a flexible one, a high bandwidth one, and so on. Once I understand this (if I can;-) I might be able to provide some suggestions.

Best

Rick

Feedback paths provide the foundation for a sociological analysis from the PCT perspective, because social interaction takes place in a shared physical environment. Feedback paths can be analyzed on a number of dimensions.

    1.      Exclusivity—Inclusivity. Is this feedback path the “private property” of one individual or is it open to use by multiple individuals? How big is the pool of potential users? How many individuals depend on it for controlling their perceptions?
    2.      Flexibility of Purpose. How many different kinds of perceptions does this feedback path support? Is it single-purpose or multi-purpose? How many orders of perception can be supported by this feedback path? How many different perceptions at

any given level does this feedback path support for the individual user?

    3.      Bandwidth. How strong is this feedback channel for maintaining control of a given perception? Feedback bandwidth may be increased by multiplying the effects of physical actions that compensate for disturbances, magnifying the quantity or quality

of information passed to perceptual channels, or protecting the channel from possible disturbances.

    4.      Fixity—Evanescence. How nearly permanent or resistant to random physical forces is this feedback channel? Does it need to be reconstructed every time a given perception is controlled?
    5.      Manipulability. How easily can this feedback channel be manipulated (changed in a controlled way) to meet the objectives of the user?

Hypothesis: An individual’s liberty and flexibility of action is increased when the individual has access to feedback paths that are exclusive, multi-purpose, high bandwidth, and fixed but highly manipulable. (Which must be why people fall in love with the
hand-held electronic devices that satisfy most of those requirements.)

Kent

Richard S. Marken PhD

rsmarken@gmail.com

www.mindreadings.com

Richard S. Marken PhD

rsmarken@gmail.com

www.mindreadings.com

bob hintz 11.05.12

I am really enjoying this exchange and second the opinion that Kent has just expressed. When we really start to get a good understanding of human control we will be able to talk about how that multitude of people were able to coordinate their control to create the tv program that an individual could choose to watch.

bob

···

On Mon, Nov 5, 2012 at 4:26 PM, McClelland, Kent MCCLEL@grinnell.edu wrote:

[From Kent McClelland (2012.11.05.1350)]

Here are some more of my thoughts about previous posts on this thread.

Rick Marken (2012.11.04.0840)

KM: In response to my examples of feedback paths that differ along the five dimensions I set out, Rick writes:

RM: Thanks, Kent. this is a good start. But my understanding is that a feedback path is a connection between system outputs and a controlled variable. So in order to get a better understanding of your feedback paths I have to know what variables you think
are being controlled and the means people use to control them. Once I know these things – controlled variables and the means available for controlling them – I think I’ll be able to understand better your classification of the feedback paths connecting
them.

And I responded. . .

KM: Backing up, the reason I’m focusing on feedback paths is because it seems like people interested in PCT have already done a pretty good job of exploring the psychological workings of the control loops inside a person’s body, looking at perceptual input
functions, comparator functions, output functions, and the various signals that connect these functions. But what makes PCT approach distinctive is that we insist that it’s also important to consider feedback functions. However, my sense of where we’re at
is that the PCT analysis of feedback functions is relatively undeveloped, other than paying close attention to the influence of disturbances.

RM: I think what makes PCT distinctive is the emphasis on the fact that behavior is organized around the control of perceptual variables (controlled variables) and that understanding behavior is really a matter of learning what these variables are and
how they are controlled. I think one has to do research aimed at the discovery of controlled variables before it makes any sense to analyze feedback functions since (as I said above) feedback functions are connections between system outputs and controlled
variables. If you don’t know what variables are being controlled you can’t possibly know which outputs are affecting them or how they are affecting them (the nature of the feedback function).

KM: In describing the properties of feedback paths, I have been talking mostly about manufactured objects and artifacts, and for many of these things their intended uses (perceptions that the user is seeking to control) are obvious to anyone who is culturally
competent. It doesn’t take an elaborate test for the controlled variable to understand the perceptions that the ordinary users of these artifacts are trying to control. And the feedback paths that I’m talking about (in other words, the artifacts in question)
are precisely the means that people are using to control these perceptions.

Let’s consider artifacts like a dishwasher and a baseball glove. Of course, a dishwashing machine or a baseball glove could potentially be used for a myriad of purposes depending on the creativity and special interests of the user, but 99 percent of the time
the dishwasher will be used to clean dirty dishes and the baseball glove to catch baseballs. If you saw someone putting frozen foods in a dishwasher and dirty dishes in a freezer, you might think that doing a test for the controlled variable was in order,
and likewise, if someone were wearing a baseball glove on his head as he tried to snag a baseball with his hat, but most manufactured objects have been designed for specific uses or a set of uses that are not too difficult to figure out (or at least not too
difficult until you read the instruction booklet that comes along with them, at which point you might be really confused!).

In your first paragraph above, you describe a feedback path as the connection between system outputs and the controlled variable in the environment, but I would have thought that a better name for that segment of a control loop would be the feedback function.
By feedback path, I’m referring to the whole segment of the control loop from the system’s output to the input, and in my view the controlled variable can be considered part of the feedback path.

This distinction between feedback function and feedback path can be an important one, particularly if the controlled variable of interest to the user of the feedback path has been brought into control largely or entirely by the efforts of other living individuals
or mechanical control systems, a condition we encounter quite frequently when we take the PCT analysis out of the psychological laboratory and attempt to apply it to social systems in the wild, as it were. Let me give an example.

Suppose you have a favorite television program, and you really like to experience the perception of yourself enjoying the kind of entertainment it provides. What can you do if you want to experience that perception again? Option A: Think back over episodes
of the program you have seen and construct an elaborate imaginary story in your head that stars the regulars on the show, then sit back and watch it unfold on your “eyelid movies.” Option B: Turn on the TV set when the program comes on (or look up an episode
on Hulu) and watch an episode that a big team of actors, directors, producers, camera operators, TV executives, makeup artists, accountants, sales execs, etc., etc., has prepared for your enjoyment. These are two different feedback paths for controlling the
same (or very similar) perceptions.

Which option requires more effort on the part of the user? Which option is likely to provide a more satisfactory experience? How much real controlling (and of what variables) does the individual in question need to do in order to experience the controlled variable
of interest–the experience of being entertained-- when it is handed to him or her as a prepackaged episode? I expect that 99 out of 100 individuals wanting to have that experience will choose option B, and I would regard the TV episode as one component of
a commonly used feedback path for satisfying that kind of desire for entertainment.

from my point of view as a sociologist, there are more interesting questions than which perceptions are people controlling when they use familiar objects in the culturally expected ways (and I’m speaking, I think, on behalf of most social scientists, if they
only understood PCT). Questions like (1) what are the consequences if lots of people choose one kind of feedback path instead of another to control a given kind of perception (as when most people drive to work instead of taking mass transit), and (2) when
we see people controlling obvious perceptions by using stereotypical feedback paths, what other perceptions may be “motivating” that behavior (or in other words, what may be the higher-level perceptions for which their obvious control of the lower-level perceptions
can be a means to an end).

Let me say again, if we want to spread the word about PCT among social scientists, we need to focus on a level of analysis that they are really interested in, and the predominant focus of PCT scholars on what happens inside the person’s head, while paying less
attention to what happens in the person’s environment, has made it harder for us to communicate with many social scientists.

Kent

On Nov 4, 2012, at 10:43 AM, Richard Marken wrote:

[From Rick Marken (2012.11.04.0840)]

Kent McClelland (2012.11.03.1445)

Rick Marken (2012.11.02.1615)

RM: It would help me if you gave some concrete examples of each of these types of feedback paths. …

KM: Sure, I could provide some examples.

  1. Exclusive vs. Inclusive feedback paths: your private, fenced-in backyard garden vs. a national park; a single-seater ultralight plane vs. a jumbo jet that carries hundreds of people.

RM: Thanks, Kent. this is a good start. But my understanding is that a feedback path is a connection between system outputs and a controlled variable. So in order to get a better understanding of your feedback paths I have to know what variables you think are
being controlled and the means people use to control them. Once I know these things – controlled variables and the means available for controlling them – I think I’ll be able to understand better your classification of the feedback paths connecting them.

KM: Backing up, the reason I’m focusing on feedback paths is because it seems like people interested in PCT have already done a pretty good job of exploring the psychological workings of the control loops inside a person’s body, looking at perceptual input
functions, comparator functions, output functions, and the various signals that connect these functions. But what makes PCT approach distinctive is that we insist that it’s also important to consider feedback functions. However, my sense of where we’re at
is that the PCT analysis of feedback functions is relatively undeveloped, other than paying close attention to the influence of disturbances.

RM: I think what makes PCT distinctive is the emphasis on the fact that behavior is organized around the control of perceptual variables (controlled variables) and that understanding behavior is really a matter of learning what these variables are and how they
are controlled. I think one has to do research aimed at the discovery of controlled variables before it makes any sense to analyze feedback functions since (as I said above) feedback functions are connections between system outputs and controlled variables.
If you don’t know what variables are being controlled you can’t possibly know which outputs are affecting them or how they are affecting them (the nature of the feedback function).

KM: Most social scientists–including most economists, political scientists, sociologists, and anthropologists–aren’t particularly interested in the psychological details of how neural connections work. Instead, they are intensely interested in understanding
and explaining empirically observable objects, patterns of human action, and events that occur outside the human body.

RM: I agree. And I think that what PCT shows is that in order to understand these things social scientists have to understand what variables people are controlling, how they control them (the lower order means used) and why (the higher order reasons for controlling
the variable at particular levels).

KM: From the point of view of PCT, it seems to me, objects, actions and events can all be analyzed in terms of feedback functions.

RM: Again, not before you find out what variables are being controlled and the outputs used to control them. Once you know those things then the nature of feedback functions would, indeed, contribute to our understanding of observable aspects of human behavior.

Best

Rick

In the context of control loops, objects, actions, and events serve to provide feedback paths for controlling perceptions. If we can focus on explaining objects, actions, and events from a PCT perspective, we’ll be meeting social scientists more than halfway.

The examples I’ve given above are all examples of objects used as feedback paths. In economic lingo, I’m talking about goods, and I think we could make a similar analysis of services: the patterns of action that serve as feedback paths for controlling the perceptions
people want to control. Events might be a little trickier to analyze, except that they often occur as a confluence of objects and actions. An election, for instance, is an event that brings together objects like buildings, voting machines, and TV sets, with
actions like debates, news reporting, and people going to the polls. If we look at how all these various things serve as feedback paths for controlling the various participants’ perceptions, we’re on our way to a PCT analysis of this complex phenomenon.

Choosing one kind of feedback path instead of another for controlling a perception can be consequential in lots of ways, and I imagine you could come up with some examples of that yourself . . .

Best,

Kent

Kent

On Nov 2, 2012, at 6:14 PM, Richard Marken wrote:

[From Rick Marken (2012.11.02.1615)]

Kent McClelland (2012.11.02.1250)]

KM: As a sociologist I’m interested in looking at the environment as consisting of possible feedback paths for controlling different kinds of perceptions. I’ve done some thinking about how one might classify feedback paths, and I’d be happy to hear any corrections,
additions, or suggestions:

RM: It would help me if you gave some concrete examples of each of these types of feedback paths. For example, I’d like a real world example of an exclusive feedback path and a non-exclusive one so I can see the difference. Right now I wouldn’t know on what
basis to conclude that a person is controlling something via an exclusive feedback path, a flexible one, a high bandwidth one, and so on. Once I understand this (if I can;-) I might be able to provide some suggestions.

Best

Rick

Feedback paths provide the foundation for a sociological analysis from the PCT perspective, because social interaction takes place in a shared physical environment. Feedback paths can be analyzed on a number of dimensions.

    1.      Exclusivity—Inclusivity. Is this feedback path the “private property” of one individual or is it open to use by multiple individuals? How big is the pool of potential users? How many individuals depend on it for controlling their perceptions?
    2.      Flexibility of Purpose. How many different kinds of perceptions does this feedback path support? Is it single-purpose or multi-purpose? How many orders of perception can be supported by this feedback path? How many different perceptions at

any given level does this feedback path support for the individual user?

    3.      Bandwidth. How strong is this feedback channel for maintaining control of a given perception? Feedback bandwidth may be increased by multiplying the effects of physical actions that compensate for disturbances, magnifying the quantity or quality

of information passed to perceptual channels, or protecting the channel from possible disturbances.

    4.      Fixity—Evanescence. How nearly permanent or resistant to random physical forces is this feedback channel? Does it need to be reconstructed every time a given perception is controlled?
    5.      Manipulability. How easily can this feedback channel be manipulated (changed in a controlled way) to meet the objectives of the user?

Hypothesis: An individual’s liberty and flexibility of action is increased when the individual has access to feedback paths that are exclusive, multi-purpose, high bandwidth, and fixed but highly manipulable. (Which must be why people fall in love with the
hand-held electronic devices that satisfy most of those requirements.)

Kent

Richard S. Marken PhD

rsmarken@gmail.com

www.mindreadings.com

Richard S. Marken PhD

rsmarken@gmail.com

www.mindreadings.com

[From Bill Powers (2012.11.06.0330 MST)]

Kent McClelland
(2012.11.05.1350)]

KM: In describing the properties of feedback paths, I have been
talking mostly about manufactured objects and artifacts, and for many of
these things their intended uses (perceptions that the user is seeking to
control) are obvious to anyone who is culturally competent. It doesn’t
take an elaborate test for the controlled variable to understand the
perceptions that the ordinary users of these artifacts are trying to
control. And the feedback paths that I’m talking about (in other words,
the artifacts in question) are precisely the means that people are using
to control these perceptions.

In place of the concept of affordance, I would simply use the term
“properties.” The properties of something are a description of
how it acts when acted upon. A hammer has mass and moment of inertia;
grasping it in a certain way and swinging it so its head contacts a nail
at a certain speed or above will result in the nail being driven into a
piece of wood. Other ways of swinging it will break a window or a cup, or
bruise a thumb. Still other ways of wielding it will send it flying
through the air to land some distance away, either destructively or
nondestructively.

So far there is no discussion of purpose, just a description of actions
and consequences.

Affordance includes purposes as well as consequences in general. But what
the hammer does and how it is wielded depend on what the intended effect
is, and that intention does not exist in the hammer. It is in the brain
of the wielder of the hammer. The affordances in the hammer do not make
the person use that hammer for any particular purpose. But now that I’ve
written that, it occurs to me to wonder if Gibson was asserting the
opposite: that it is something about the hammer that results in our using
it in a certain way. There is a strong flavor of causality in the way
Gibson described affordances: the affordances tell us what to do with the
object.

But isn’t that the main message of classical behaviorism? In behaviorism,
it is not an inner purpose that guides behavior toward achievement
of a preselected result; the behavior is shaped by the effects it has on
the world outside and thus on the occurrance of reinforcements. In
behaviorism, behavior is shaped by the affordances in the
environment.

If we remove affordances from the environment, and instead speak of
properties of the environment and purposes of the organism, we are back
to the PCT view of behavior. The organism learns what it has to do to the
environment in order to get some intended result back from it. The
intention comes first. There are generally many ways to act that will
have the same desired result, some ways requiring less effort than others
or having more reliable effects. As the organism learns it may try many
different ways. A child may hold a hammer by its head and try bumping the
head against a nail, probably in vain. It takes a while to acquire the
lower-order control systems that will make the best use of a tool’s
properties. The designer of a tool may shape it to make one way of using
it particularly effective, but the user of the tool must still discover
or otherwise learn what that way is and how to carry it out. And of
course the user must have a result in mind to use the tool at all, and
that result may not be the one the designer of the tool was thinking of.
A claw hammer, I have found, is an excellent tool for propping a basement
window open, but that use doesn’t require holding the handle near its end
and swinging the head against something.

Affordance conflates intention and consequence. I suiggest that we would
be better off to keep those ideas separate.

Best,

Bill P.

[From Chad Green (2012.11.06.1200)]

Bob, I prefer this scenario:

"When we really start to get a good understanding of human
control we will be able to talk about how [individuals] were
able to coordinate their control to create the tv program that
[a multitude of people] could choose to watch."

Imagine as many channels as there are individuals, each with its own ever-shifting storyline, making sense of their daily lives through the vicarious experiences of others.

Chad

Chad Green, PMP
Program Analyst
Loudoun County Public Schools
21000 Education Court
Ashburn, VA 20148
Voice: 571-252-1486
Fax: 571-252-1633

"If you want sense, you'll have to make it yourself." - Norton Juster

Bob Hintz <bob.hintz@GMAIL.COM> 11/5/2012 11:48 PM >>>

bob hintz 11.05.12

I am really enjoying this exchange and second the opinion that Kent has
just expressed. When we really start to get a good understanding of human
control we will be able to talk about how that multitude of people were
able to coordinate their control to create the tv program that an
individual could choose to watch.

bob

[From Kent McClelland (2012.11.07.1445 CST)]

I’ve been working my way through responses to the accumulated posts in this thread, but I’m going to jump immediately to Bill’s most recent post, because it offers a chance to make a point that I’ve been wanting to make about feedback paths.

I agree with what Bill is saying about affordances in the post below. He’s absolutely right, in my view, that the term is unhelpful, and, like him, I would prefer to not use that terminology. But I agree only up to a point with the rest of his post. I think
that to communicate our message effectively to social scientists we will need a far more sophisticated view of feedback than has usually been case in PCT discussions, as I will explain below. First, here is what Bill says:

[From Bill Powers (2012.11.06.0330 MST)]

In place of the concept of affordance, I would simply use the term “properties.” The properties of something are a description of how it acts when acted upon. A hammer has mass and moment of inertia; grasping it in a certain way and swinging it so its head
contacts a nail at a certain speed or above will result in the nail being driven into a piece of wood. Other ways of swinging it will break a window or a cup, or bruise a thumb. Still other ways of wielding it will send it flying through the air to land some
distance away, either destructively or nondestructively.

So far there is no discussion of purpose, just a description of actions and consequences.

Affordance includes purposes as well as consequences in general. But what the hammer does and how it is wielded depend on what the intended effect is, and that intention does not exist in the hammer. It is in the brain of the wielder of the hammer. The affordances
in the hammer do not make the person use that hammer for any particular purpose. But now that I’ve written that, it occurs to me to wonder if Gibson was asserting the opposite: that it is something about the hammer that results in our using it in a certain
way. There is a strong flavor of causality in the way Gibson described affordances: the affordances tell us what to do with the object.

But isn’t that the main message of classical behaviorism? In behaviorism, it is not an inner purpose that guides behavior toward achievement of a preselected result; the behavior is shaped by the effects it has on the world outside and thus on the occurrance
of reinforcements. In behaviorism, behavior is shaped by the affordances in the environment.

If we remove affordances from the environment, and instead speak of properties of the environment and purposes of the organism, we are back to the PCT view of behavior. The organism learns what it has to do to the environment in order to get some intended result
back from it. The intention comes first. There are generally many ways to act that will have the same desired result, some ways requiring less effort than others or having more reliable effects. As the organism learns it may try many different ways. A child
may hold a hammer by its head and try bumping the head against a nail, probably in vain. It takes a while to acquire the lower-order control systems that will make the best use of a tool’s properties. The designer of a tool may shape it to make one way of
using it particularly effective, but the user of the tool must still discover or otherwise learn what that way is and how to carry it out. And of course the user must have a result in mind to use the tool at all, and that result may not be the one the designer
of the tool was thinking of. A claw hammer, I have found, is an excellent tool for propping a basement window open, but that use doesn’t require holding the handle near its end and swinging the head against something.

Affordance conflates intention and consequence. I suiggest that we would be better off to keep those ideas separate.

KM: The hammer that you’re talking about, Bill, is not just an object with physical properties. It is a manufactured artifact that has been designed with socially standardized physical properties. There are, no doubt, millions of claw hammers sitting in
people’s basements and toolboxes all over America and around the world, and I would guess that 99 percent of them look pretty much the same, with a standard length of handle, a standard shape of head, and a standard weight (usually 16 ounces in the USA). Of
course, the hammers will have been produced by various manufacturers, and the material composition of the hammers may also vary, with some having metal handles covered with rubber, for instance, while others have wooden handles. In almost every case the head
of the hammer will be made of metal, typically steel. But any variations in the fundamental design of the hammer–its length, distribution of weight, and heft–will occur only within extremely narrow parameters. There’s no law that says that hammers must be
constructed with a given shape and weight, but a complex history of the traditional shaping of the tool to match its customary uses by craftsmen, followed by market competition between manufacturers of hammers to produce tools matching that customary design,
has ultimately led to a product that is largely standardized.

This standardization of products like hammers has some important consequences from a PCT perspective. First, the person who has learned how to use a hammer (in precisely the trial-and-error-and-reorganization way that Bill describes in his post) can pick
up another hammer and immediately be able to use it in much the same way as previously used hammers. Of course, the feel of the new hammer might be slightly different, but the muscular reorganization process necessary to use the hammer efficiently (to hammer
in nails, of course) can proceed very rapidly, because the new hammer is so much like previous ones.

Second, while some users of hammers will find creative and unusual uses for the tool, as in the window-propping example Bill gives, a social scientific observer studying the uses of hammers would be likely to observe that most users most of the time use
hammers in the conventionally expected way to pound in nails to fasten pieces of wood together. And because imitation is an important part of the human learning process, neophytes learning to use hammers are likely to imitate experienced users of hammer by
using the tool (most of the time) in a way that is similar, if not precisely identical, to the way that the great majority of other users have traditionally used it. In other words, there is a lot of cultural predictability (if not total uniformity) in the
ways that hammers are used. Thus, relative uniformity in the physical environment–the standardized design of the hammer as an artifact–goes hand in hand (or hammer in hand?) with relative uniformities in behavioral patterns.

Third, and most importantly, the hammer exemplifies an important fact about the living environments of people in economically developed countries. We are surrounded by manufactured artifacts that have been designed, produced, and marketed with culturally
conventional uses in mind. Virtually everything that we see or touch in our homes and workplaces is some kind of manufactured artifact that has been carefully designed to help us control one perception or another. Purposes, in the form of the intended uses
of manufactured objects, have in some sense been built into the myriad objects that surround us.

Objects like hammers can be seen as controlled environmental variables that result from an extremely complex collective control process in which “a multitude of people,” to use Bob Hintz’s words, (the miners, mine owners, foremen, transport workers, transport
company owners and executives, manufacturing executives and owners and stockholders, factory workers, engineers, designers, sales workers, accountants, OSHA regulators, more transport workers, wholesale distributors, yet more transport workers, advertisers,
retail store sales and management workers, and probably many more people that I’ve overlooked) have cooperated in setting up and maintaining the complicated supply, manufacturing, and distribution chains that bring hammers to market and put them into the hands
of the end users.

From the perspective of the user, the hammer is not in itself a controlled environmental variable, except as its motions are controlled by the user to achieve control of the perception of using nails to fasten pieces of wood together. (Notice that 2-by-4’s
and the nails used to fasten them together are also standardized products with their intended uses also in some sense built in.) The hammer, nails, and pieces of wood are simply the means chosen by the user to control the perception of building a planned wooden
structure, and not ends in themselves. In the PCT analysis that I am trying to develop, they serve as segments of a feedback path to link the user’s physical actions with the perception that the user is seeking to control of successfully building the planned
structure.

As (part of) the user’s feedback path, the hammer (1) is an item designed for exclusive use of one person at a time, that (2) has somewhat limited flexibility for helping to control perceptions other than hammering in nails, but nonetheless (3) serves
as a pretty effective means for controlling that perception, and (4) is a durable and thus relatively fixed object that (5) is pretty difficult (but probably not impossible) for the user to modify to suit his own requirements.

Considering this complex social and cultural context for the hammer as an object, Bill is right that it would be naive and oversimplified to describe the hammer as offering “affordances,” with its purposes somehow emerging from its physical qualities,
as if it were some natural object to be picked up and found a use for, like a rock that a stone-age person might pick up and then try bashing things with it. On the contrary, the hammer has been designed, shaped, and distributed by a collection of people who
had a specific set of purposes in mind–a tool for sinking nails and pulling them out again. That it works well for those purposes is fully intentional, not an accident of nature.

By the same token, it seems to me a little naive and oversimplified for PCT analysts to view the hammer simply as an object with physical properties, ignoring all the social and cultural baggage carried with it–the many people’s purposes that its form
embodies. To communicate effectively with social scientists about things like hammers, PCT advocates will need to get more sophisticated in describing how large collections of people can provide standardized feedback paths for each other to use in controlling
a multiplicity of individual perceptions.

Enough ranting for today. . . I’m probably going on and on because I stayed up until 2 AM last night to watch Romney’s hastily written concession speech and Obama’s highly polished acceptance speech (a speech that was well worth waiting up for!).

Look for another installment on the subject of feedback paths in a day or two.

My best to all,

Kent

PS to the sociologically inclined: Two interesting recent books by sociologists describe the production and enforcement of standards, a topic closely related to the collective control processes that I have described above. I take these books as hopeful
signs that my field is finally getting ready to accept a PCT analysis of social structure and culture, which could put all these concerns into a more coherent context than sociologists have been able to do so far.

Busch, Lawrence. 2011. Standards: Recipes for Reality. Cambridge, MA: MIT Press.

Lampland, Martha and Susan Leigh Star. 2009. Standards and Their Stories: How Quantifying, Classifying, and Formalizing Practices Shape Everyday Life. Ithaca, NU: Cornel University Press.