What's in a name (was Analyzing Feedback Paths)

[Martin Taylor 2012.11.28.23.11]

[From Rick Marken (2012.11.25.0940)]

Martin Taylor (2012.11.25.00.23)--

RM: I don't see what changing minds has to do with it. It's just a
logic problem, as far as I can tell. Martin says X is true. Bill and I
say X is false.

MT: Could you point to the place where this happened?

Let X = *ffordance. My overall impression has been that Bill and I
have been saying things that show that X is either wrong, misleading
or useless, ie. false.

Have you actually read anything I have written about effordances? If you answer "yes", can you explain how "effordance" can in itself be wrong? To quote the most succinct description of an effordance from previous messages, an effordance "is just a condition in a person's environment through which a person can control a perception." To call such a thing "wrong" is a categorical error. A condition in a person's environment is neither right nor wrong. It just is.

I grant that Bill has been going to some length to show that "affordances" may mislead people into thinking that external conditions induce people to perform some actions. It was in agreement with Bill on this that I disagreed with Kent, and it is why I coined the word "effordance" as a substitute for "environmental affordance". I cannot help it if people are misled by the use of a neologism, any more than Bill can help it if people learning PCT are misled by thinking a "perception" must be consciously experienced. At least I am using a new word, rather than using an old word in a new sense.

As for whether the concept is useless, all effordances are useless if nobody uses them to control a perception. For you, the concept of "effordance" is useless because you are not controlling a perception for which the concept might be useful. That doesn't mean it is useless for others who are interested in how the control behaviour of one person may affect the possibilities for control by another person -- which was Kent's original issue.

I'm not sure how "i.e." (= "id est", which means "that is") follows from "wrong, misleading, or useless", nor do I follow how "a condition in a person's environment through which a person can control a perception" can be false, unless by trying to use it the person fails to influence the perception in question.

MT: So far as I can see...You were apparently agreeing with me.

Quote [From Rick Marken (2012.11.1710)]:

What is
being insisted, I think, is that what an environmental object "makes
available" (affords) depends on the purposes of the person using that
object. What is being insisted, in other words, is that you have to
take_controlled variables_ into account before you can have any idea
what any particular "object" in the environment "makes available."

MT: That is, and has been, my position.

RM: Then you agree with Bill and I that the concept of *ffordance is
unnecessary because it refers to things that are already included in
PCT as disturbances or feedback functions.

You truly haven't read anything I have written about effordances. That being the case, I see little point in writing more about them in dispute with you. However, before moving on to using the concept in controlling scientific perceptions I am trying to control, I will point out that a feedback function is not an effordance, though the feedback function for a specific control loop is influenced by what effordance(s) is/are used in the environmental feedback path.

Please note for future reference that the concept of effordance is unrelated to the concept of disturbance apart from the fact that both belong to the set of concepts that relate to environmental effects in control.

If I thought you were trying to understand rather than simply trying to oppose for opposition's sake, I would pursue the argument further. But what's the point?

Martin

[Martin Taylor 2012.11.29.09.43]

[From Bill Powers (2012.11.25.1755 MDT)]

The coin game is a good example to use in speaking of affordances (Taylor's effordances, meaning properties of the environmental feedback function). There are huge numbers of effordances for any controlled variable -- actually infinite numbers of them when we think quantitatively. Many quantitatively different effordances make no important difference in controlling a given controlled variable, since control systems can handle wide variations in the loop gain without materially altering the quality of control.

I think rick Marken is right when he keeps insisting that you can't even define an *ffordance until you know what the controlled variable is. It's only when one thinks qualitatively, in categories rather than quantities, that the concept makes any degree of sense. And then that degree seems pretty small.

There are two key points in "you can't even define an *ffordance until you know what the controlled variable is". The first is the word "an", which makes it clear you are not talking about the general case (the "x" in an algebraic equation) but a particular case (the "3" in an arithmetic sum). In my writings, I think I have tried to make the truth of "you can't even define an *ffordance until you know what the controlled variable is" clear. That is correct, but to study Kent's propositions one does not need to define any particular effordance. Kent suggests, for example, that the nature of the controlled perception is irrelevant to the question of whether one has more freedom to control if an effordance used in control is permanently available than if it comes and goes erratically.

The second issue with the quote is the "*" in "*ffordance", which suggests that there is no distinction between a Gibsonian "affordance" in which the environment imposes upon the organism and an "effordance" which the organism selects from the environment. Throughout these threads, there has been a failure to distinguish the two concepts, a failure for which I must accept some of the blame, since I used the word "affordance" in defining "environmental affordance". I did that because the everyday concept of "afford" does carry the right connotation when you are thinking PCT, and because I could think of no more appropriate label. More recently, I thought that replacing the "a-" in affordance with "e-" would resolve the issue, but you keep replacing the "e-" with a "*-" as though it made no difference. So long as you keep thinking that there is no difference, I suspect that you will continue to think that the whole concept is pretty useless. Please, if you are talking of affordances, use the "a-", and if you are talking of effordances, use the "e-". That way, it's easier to know which you intend.

Now we come to the quantitative question, which is a direction I had intended to take the discussion of effordance in social situations. An effordance is the fact that one can influence a controlled perception by manipulating something in the environment. In the coin game, one such "something" may be the location of a coin on the table. The effordance is not the actual location, nor is it the amount by which a person moves the coin. It is the fact that by moving the coin the person influences the controlled variable, whether that controlled variable be "keeping the coins in a straight line" or "making the product of the distances between two coins in centimeters and their values in cents be equal to some reference value". Either of these very different controlled perceptions can be influenced by moving the coin.

Bill says: "Many quantitatively different effordances make no important difference in controlling a given controlled variable, since control systems can handle wide variations in the loop gain without materially altering the quality of control." This is quite true, but the converse is also true, that differences in loop gain for a given transport lag can make appreciable differences in the quality of control, even to the point of taking the loop into oscillation, or conversely, making a previously unstable loop stable. But I'm not at all clear how a "quantitatively different effordance" would apply to the coin game, unless it refers to the stickiness of the coin-table interface that would make it more or less difficult to move the coin. If that's what you mean, a stickier interface would probably slow the control loop or make the movements more jerky. Neither effect would be problematic if the controller had infinite time to bring the perception to a stable reference value in the absence of external disturbance, but they would influence the precision of control in the presence of a changing disturbance or reference value. A change of effordance usually causes a change in the parameters of the environmental feedback function, but is unrelated to the values of the variables at any point in the loop.

An instance of how a change of effordance can cause a change in the parameters of the environmental feedback function can be seen in the "crossing the brook" example. So far in discussion of the example we have assumed a simple plank strong enough and placed sufficiently stably to allow a person to walk confidently across it. But now suppose the ends are on unstable ground, or the plank is thin, or it is a tree that has fallen across the stream, with varying degrees of slippery moss cover. All of these influence the ease with which the effordance can be used in control of the perception of walking across the brook. Assuming that none of these conditions remove the effordance entirely (make it impossible for the person to use the object to cross the brook), yet there may come a point where the person is likely to choose to walk upstream to a more formal bridge rather than cross using with difficulty the object lying across the brook.

One cannot really discuss effordances and Kent's hypotheses seriously without considering reorganization. I want to get into that discussion, but it is a different discussion from the issues I am trying to address here, which relate simply to the nature of an effordance and what quantitative parameters of an effordance must be considered (and even that last is closely tied to reorganization, because a person's skill determines whether and how easily a particular environmental condition can be used in controlling a particular perception).

Martin

[From Rick Marken (2012.11.29.0800)]

Martin Taylor (2012.11.28.23.11)–

RM: Let X = *ffordance. My overall impression has been that Bill and I

have been saying things that show that X is either wrong, misleading

or useless, ie. false.

MT: Have you actually read anything I have written about effordances? If you answer “yes”, can you explain how “effordance” can in itself be wrong? To quote the most succinct description of an effordance from previous messages, an effordance “is just a condition in a person’s environment through which a person can control a perception.” To call such a thing “wrong” is a categorical error. A condition in a person’s environment is neither right nor wrong. It just is.

Let’s try this: Given the definition of effordance as “just a condition in a person’s environment through which a person can control a perception” can you point to anything in the environment that is not an an effordance?

Best

Rick

[Martin Taylor 2012.11.29.15.13]

You can't point to anything in the environment and say it IS an

effordance. In itself, nothing is. You have to include the
perception to be controlled and the actions. The plank lying across
the brook is not an effordance (at least not one that many
reorganizations would discover) for perceiving yourself to be more
clean-shaven or for perceiving yourself to be in the middle of the
driving lane or for paying for groceries. The plank COULD BE an
effordance for many things, particularly if you have the ability and
strength to move it. But it IS NOT an effordance in the abstract.
Now you might ask if there is anything in the environment that could
NOT be an effordance. I don’t know, but I am guessing that the
answer depends on whether there is currently any means to act upon
the thing and for the action to influence some perception that might
be controlled. Even a neutrino could qualify nowadays, perhaps not
individually, but as a member of a swarm, since neutrinos can now be
detected and the result of their detection is that some perceptions
are altered, and it is possible to create a setup so that a neutrino
beam is aimed at a specific place. But neutrinos were presumably
just as abundant a century ago, and at that time they were neither
detectable nor manipulable, so I would guess that a century ago
neutrinos would provide an example of things in the environment that
could not be an effordance. There is probably a lot going on in our environment that has much
the same status now as neutrinos did a century ago. A century or a
millennium form now, those things might be available for use in
control, but they aren’t right now, so they are in the environment
but cannot be effordances for any control system. But if you can see
something and act so as to use it, it probably could be an
effordance for controlling some perception or other. Probably a lot
of different perceptions.
Looking at the question from the point of view of the condition in
the environment isn’t really helpful, most of the time. As I have
been agreeing with you, for any specific case you have to know what
perception is being controlled and what actions are being used to
influence that perception through the effordance in the environment.
Also, although it may sometimes be helpful in discovering the
parameters of the environmental feedback function to look at which
of the available effordances is being used, most of the time you
probably wouldn’t want to bother. You would want to bother if you wanted to help someone control some
variable (by saying something like “if you hold the framingulus the
other way up it might work better”) or to make controlling some
variable more difficult (by taking the plank away from the brook and
destroying the footbridge). But here we are getting into the social
area in which we might consider Kent’s hypotheses. Do you want to do
that?
Martin

···

On 2012/11/29 10:59 AM, Richard Marken
wrote:

[From Rick Marken (2012.11.29.0800)]

    Martin Taylor

(2012.11.28.23.11)–

      RM: Let X =

*ffordance. My overall impression has been that Bill and I

      have been saying things that show that X is either wrong,

misleading

      or useless, ie. false.
    MT: Have you actually read anything I have written about

effordances? If you answer “yes”, can you explain how
“effordance” can in itself be wrong? To quote the most succinct
description of an effordance from previous messages, an
effordance “is just a condition in a person’s environment
through which a person can control a perception.” To call such a
thing “wrong” is a categorical error. A condition in a person’s
environment is neither right nor wrong. It just is.

  Let's try this: Given the definition of effordance as "just a

condition in a person’s environment through which a person can
control a perception" can you point to anything in the environment
that is not an an effordance?

[From Rick Marken (2012.11.29.1445)]

Martin Taylor (2012.11.29.15.13)--

RM: Let's try this: Given the definition of effordance as "just a condition in a
person's environment through which a person can control a perception" can
you point to anything in the environment that is _not_ an an effordance?

MT: You can't point to anything in the environment and say it IS an effordance.
In itself, nothing is. You have to include the perception to be controlled
and the actions.

RM: So it sounds like effordance is just an ugly way of saying
feedback function; as I said, it doesn't exist until you know what
variables are being controlled.

MT: Now you might ask if there is anything in the environment that could NOT be
an effordance.

RM: Actually, I did ask precisely that;-)

MT: I don't know, but I am guessing that the answer depends on
whether there is currently any means to act upon the thing and for the
action to influence some perception that might be controlled.

RM: But if there isn't currently any means of controlling the
perception then the means doesn't exist (such a rocket that would take
people to the moon and back in 1960). And if the means does not exist
it doesn't exist in the environment. So it's not an effordance by your
definition which says it's "a condition in a person's environment".

MT: Even a
neutrino could qualify nowadays, perhaps not individually, but as a member
of a swarm, since neutrinos can now be detected and the result of their
detection is that some perceptions are altered, and it is possible to create
a setup so that a neutrino beam is aimed at a specific place. But neutrinos
were presumably just as abundant a century ago, and at that time they were
neither detectable nor manipulable, so I would guess that a century ago
neutrinos would provide an example of things in the environment that could
not be an effordance.

RM: Ignoring for the moment that a neutrino is a theoretical concept
and allowing the possibility that it is an actual environmental
entity, it was, by your definition, an effordance even a century ago,
since was (and is) a "condition in a person's environment through
which a person can control a perception". If neutrinos were not
effordances when people didn't know how to use them to make neutrino
beams then Excel is not an effordance for people who don't know how to
use it to balance the books.

MT: There is probably a lot going on in our environment that has much the same
status now as neutrinos did a century ago.

RM: Right, and, more to the point, there are organizations of these
environmental variables that could help us control things better once
we figure out how to organize them properly; it's called invention.

MT: You would want to bother [with effordances] if you wanted to help
someone control some variable (by saying something like "if you hold
the framingulus the other way up it might work better") or to make
controlling some variable more difficult (by taking the plank away
from the brook and destroying the footbridge).

RM: This just sounds like suggesting variables to control as the means
of controlling other variables. I think any student of PCT would
understand that suggesting lower level perceptions to control as the
means of controlling higher level ones would be a fundamental
component of teaching people to control. I don't see what effordance
adds.

MT: But here we are
getting into the social area in which we might consider Kent's
hypotheses. Do you want to do that?

RM: By all means. I have been trying to figure out what the science
(sociology) of effordance would be about. So how about describing some
data and showing me how the concept of effordance can contribute to
our understanding of what's going on.

Best

Rick

···

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

[From Erling Jorgensen (2012.11.29.1800)

Rick Marken (2012.11.29.0800)

Martin Taylor (2012.11.28.23.11)--

To quote the most succinct description of an effordance from previous
messages, an effordance "is just a condition in a person's environment
through which a person can control a perception." To call such a thing
"wrong" is a categorical error. A condition in a person's environment is
neither right nor wrong. It just is.

Let's try this: Given the definition of effordance as "just a condition in
a person's environment through which a person can control a perception" can
you point to anything in the environment that is _not_ an an effordance?

I tend to agree with Rick, here. When we end up saying in effect that
'anything can be something', it may not be that helpful a concept.

Rather than looking at the supposed properties of objects in the
environment, it seems that the key aspect that Kent has been asking about
is the effects of stabilization in the environment. I am wondering if
the notion of 'niches of stabilization' could advance the discussion.

It seems to me that control systems not only counter disturbances, & thus
create perceptual stabilities. They also exploit stabilities that they
find in the environment. I guess I'm talking about a network of control
systems such as an organism doing the exploiting here. There is a
certain impartiality about it, in that it really doesn't matter from
where those stabilities arise, whether from physics or inorganic means
or from the side effects of other control systems creating their own
perceptual stabilities.

Control systems are opportunistic. They'll use whatever works to bring
their own perceptions under control. Because of that, they appear to be
tailor-made to slip into whatever niches of stabilization that they come
upon in their environment.

This is really just a corollary of the whole hierarchical nature of the
proposed HPCT system. The effective 'environment' for any given level of
perceptual control includes all those (descending) implementing outputs &
(ascending) constituting perceptions at lower levels of control. Higher
levels seem fairly indifferent to which particular muscle fibers help to
stabilize the control they care about. If one channel is not working,
recruit another. And it may also be the case that different collections of
weightings of the lower level perceptions may be equally successful in
constructing the higher level perception of note.

Another way to make this point is to note that any given control task is
always partitioned into lower, finer-grained but faster-operating, layers
of perceptual control. And the resulting perceptual stability migrates
upward in the hierarchy.

So then, if some of that stability is to be found in the so-called physical
environment that is consensually shared among other organisms, so be it.
If some of those organisms are themselves the source of the stability,
so be it. Feedback paths for control systems can become exceedingly
complex, passing through the actions of other living control systems, as
the early experiments by Tom Bourbon & his students seemed to demonstrate.

I propose that we think about niches of stabilization, as we consider the
sociological realm of interacting living control systems. Many of these
niches are actively created by those other control systems, while other
niches may arise more by way of side effect. Does this seem like a helpful
formulation to others?

All the best,
Erling

[From Kent McClelland (2012.11.29.2045)]

Erling Jorgensen (2012.11.29.1800)

Rather than looking at the supposed properties of objects in the
environment, it seems that the key aspect that Kent has been asking about
is the effects of stabilization in the environment. I am wondering if
the notion of 'niches of stabilization' could advance the discussion.

Yes, thank you, Erling. "Niches of stabilization" captures the idea I've been driving at quite nicely. I don't have time to comment further at the moment, but your post seems right on target to me.

Best,

Kent

···

On Nov 29, 2012, at 6:00 PM, Erling Jorgensen wrote:

[From Erling Jorgensen (2012.11.29.1800)

Rick Marken (2012.11.29.0800)

Martin Taylor (2012.11.28.23.11)--

To quote the most succinct description of an effordance from previous
messages, an effordance "is just a condition in a person's environment
through which a person can control a perception." To call such a thing
"wrong" is a categorical error. A condition in a person's environment is
neither right nor wrong. It just is.

Let's try this: Given the definition of effordance as "just a condition in
a person's environment through which a person can control a perception" can
you point to anything in the environment that is _not_ an an effordance?

I tend to agree with Rick, here. When we end up saying in effect that
'anything can be something', it may not be that helpful a concept.

Rather than looking at the supposed properties of objects in the
environment, it seems that the key aspect that Kent has been asking about
is the effects of stabilization in the environment. I am wondering if
the notion of 'niches of stabilization' could advance the discussion.

It seems to me that control systems not only counter disturbances, & thus
create perceptual stabilities. They also exploit stabilities that they
find in the environment. I guess I'm talking about a network of control
systems such as an organism doing the exploiting here. There is a
certain impartiality about it, in that it really doesn't matter from
where those stabilities arise, whether from physics or inorganic means
or from the side effects of other control systems creating their own
perceptual stabilities.

Control systems are opportunistic. They'll use whatever works to bring
their own perceptions under control. Because of that, they appear to be
tailor-made to slip into whatever niches of stabilization that they come
upon in their environment.

This is really just a corollary of the whole hierarchical nature of the
proposed HPCT system. The effective 'environment' for any given level of
perceptual control includes all those (descending) implementing outputs &
(ascending) constituting perceptions at lower levels of control. Higher
levels seem fairly indifferent to which particular muscle fibers help to
stabilize the control they care about. If one channel is not working,
recruit another. And it may also be the case that different collections of
weightings of the lower level perceptions may be equally successful in
constructing the higher level perception of note.

Another way to make this point is to note that any given control task is
always partitioned into lower, finer-grained but faster-operating, layers
of perceptual control. And the resulting perceptual stability migrates
upward in the hierarchy.

So then, if some of that stability is to be found in the so-called physical
environment that is consensually shared among other organisms, so be it.
If some of those organisms are themselves the source of the stability,
so be it. Feedback paths for control systems can become exceedingly
complex, passing through the actions of other living control systems, as
the early experiments by Tom Bourbon & his students seemed to demonstrate.

I propose that we think about niches of stabilization, as we consider the
sociological realm of interacting living control systems. Many of these
niches are actively created by those other control systems, while other
niches may arise more by way of side effect. Does this seem like a helpful
formulation to others?

All the best,
Erling

[Martin Taylor 2012.11.29.23.38]

[From Erling Jorgensen (2012.11.29.1800)

Rick Marken (2012.11.29.0800)

Martin Taylor (2012.11.28.23.11)--
To quote the most succinct description of an effordance from previous
messages, an effordance "is just a condition in a person's environment
through which a person can control a perception." To call such a thing
"wrong" is a categorical error. A condition in a person's environment is
neither right nor wrong. It just is.

Let's try this: Given the definition of effordance as "just a condition in
a person's environment through which a person can control a perception" can
you point to anything in the environment that is _not_ an an effordance?

I tend to agree with Rick, here. When we end up saying in effect that
'anything can be something', it may not be that helpful a concept.

As you may have subsequently noted from [Martin Taylor 2012.11.29.15.13], this is not the case when talking about effordances, unless it is Rick or Bill talking.

Rather than looking at the supposed properties of objects in the
environment, it seems that the key aspect that Kent has been asking about
is the effects of stabilization in the environment. I am wondering if
the notion of 'niches of stabilization' could advance the discussion.

It seems to me that control systems not only counter disturbances, & thus
create perceptual stabilities. They also exploit stabilities that they
find in the environment. I guess I'm talking about a network of control
systems such as an organism doing the exploiting here. There is a
certain impartiality about it, in that it really doesn't matter from
where those stabilities arise, whether from physics or inorganic means
or from the side effects of other control systems creating their own
perceptual stabilities.

Control systems are opportunistic. They'll use whatever works to bring
their own perceptions under control. Because of that, they appear to be
tailor-made to slip into whatever niches of stabilization that they come
upon in their environment.

This is really just a corollary of the whole hierarchical nature of the
proposed HPCT system. The effective 'environment' for any given level of
perceptual control includes all those (descending) implementing outputs &
(ascending) constituting perceptions at lower levels of control. Higher
levels seem fairly indifferent to which particular muscle fibers help to
stabilize the control they care about. If one channel is not working,
recruit another. And it may also be the case that different collections of
weightings of the lower level perceptions may be equally successful in
constructing the higher level perception of note.

Another way to make this point is to note that any given control task is
always partitioned into lower, finer-grained but faster-operating, layers
of perceptual control. And the resulting perceptual stability migrates
upward in the hierarchy.

So then, if some of that stability is to be found in the so-called physical
environment that is consensually shared among other organisms, so be it.
If some of those organisms are themselves the source of the stability,
so be it. Feedback paths for control systems can become exceedingly
complex, passing through the actions of other living control systems, as
the early experiments by Tom Bourbon & his students seemed to demonstrate.

I propose that we think about niches of stabilization, as we consider the
sociological realm of interacting living control systems. Many of these
niches are actively created by those other control systems, while other
niches may arise more by way of side effect. Does this seem like a helpful
formulation to others?

Yes. I treated much of that material a bit more mathematically in a talk given in 1997 to the University of Toronto Mathematics Club (reconstructed at <http://www.mmtaylor.net/PCT/Mutuality/index.html&gt;\). The presentation may be a bit dated, since the talk was given 15 years ago, but on looking at it just now I don't think there is much to change except on the very last page, where the conclusion "cooperation will win in the end" seems to be over-optimistic, oscillation and chaos being other long-term possibilities. The talk does treat pretty much of what you suggest above and more besides.

Martin

[Martin Taylor 2012.11.29.23.52]

No, no. It is NOT, repeat NOT, the same as the feedback function. A

feedback function is a relationship between an output variable and
an input variable. It can be written i = f(o, t). An effordance has
characteristics that affect the form of the feedback function, but
is not the function. Not precisely that. You asked: “can
you point to anything in the environment that is an an effordance?”. There’s
a big difference between “could not be” and “is not”.
At that time there was indeed nothing in the environment that
provided an effordance for getting people to the moon and back. Nor
is there now. For a while after 1969 there was. I’m not sure what
your comment is intended to illustrate or question.
No it wasn’t, because a century ago nobody could manipulate
neutrinos even to the limited extent that is now possible to a few
people, and nobody could perceive their effects on controlled
perceptions even if they had unknowingly manipulated them.
Both of those statements are true (apart from the wording that
suggests objects “are” effordances. Excel does indeed not provide an
effordance ( not “is not an effordance”) for people to control
perceptions if they don’ have the skill to use it to control those
perceptions. Neutrino beams don’t provide any effordances for almost
everyone in the world right now, though it is possible that in
centuries to come they may do.
For once, we agree. Or seem to.
But do you see where this simple observation leads when considering
the standard HPCT hierarchic organization? It’s a bit of a can of
worms, isn’t it? Doesn’t it imply that there exist control systems
that control perceptions of the magnitude of the error signal in
other control systems? I grant that the reorganizing system is
specified as doing exactly that, but construction of tools does not
seem like random reorganization, even when you watch crows,
elephants, or chimps do it, let alone when you introspect on
yourself doing it. Maybe it is, and you simply don’t see the
imagined but not executed “wrong direction” attempts at control, but
it feels as though one perceives that some system controlling a
perception is in a sustained error condition and that changing X in
the environment (moving the plank from lying on the bank to lying
across the brook) would allow that perception to be controlled.
I think you are looking at apples and calling them oranges because
they are both reasonably round objects of similar size. It’s
disingenuous.
Data follows theory or may guide toward theory, but it’s the theory
that is paramount. I refer you back to
[Martin Taylor 2012.11.25.00.23]
and
[Martin Taylor 2012.11.25.00.23]
for some relevant comments that I won’t bother repeating. Is it
“data” to say that of all the prisoners legally jailed, not one has
had a holiday on a beach in Samoa during the jail term without
escaping custody illegally. That’s 100% accuracy for a statement
that control is impossible if the requisite effordances are removed
by the action of other control systems (assuming that some of the
prisoners would have liked to have a holiday on a beach in Samoa).
Is it “data”?
I find it odd to talk about a “science of effordance”. Effordance is
a linking concept between a (presumed to exist) real world and the
abstract world of feedback functions. It isn’t a science or even a
theory. You may be able to abstract some parameters of an effordance
and use them to deduce the form of a feedback function if you are
lucky. But it’s more probable that you will be able to make
predictions, such as that a person is likely to cross the brook more
slowly over a slippery mossy log than over a sturdy solidly placed
plank. I’m not clear how you would put that into a feedback function
other than to say that its transport lag is greater if the log
provides the effordance than if the effordance is provided by the
plank.
I think you tend to ask the wrong questions, questions based around
the ideas that effordances reside in objects as properties that can
be measured in the object. They don’t. If a person has not the skill
to use an object to influence a perception, that object does not
provide an effordance for that person to control that perception.
For controlling that perception, the object might as well not exist.
You simply can’t talk about Excel providing an effordance for
balancing the books without at the same time saying “for a person
who wants to balance books and has the skill and the computational
support to use Excel.”
Martin

···
[From Rick Marken (2012.11.29.1445)]
Martin Taylor (2012.11.29.15.13)--



RM: Let's try this: Given the definition of effordance as "just a condition in a
person's environment through which a person can control a perception" can
you point to anything in the environment that is _not_ an an effordance?
MT: You can't point to anything in the environment and say it IS an effordance.
In itself, nothing is. You have to include the perception to be controlled
and the actions.
RM: So it sounds like effordance is just an ugly way of saying
feedback function; as I said, it doesn't exist until you know what
variables are being controlled.

MT: Now you might ask if there is anything in the environment that could NOT be
an effordance.
RM: Actually, I did ask precisely that;-)

not


MT: I don't know, but I am guessing that the answer depends on
whether there is currently any means to act upon the thing and for the
action to influence some perception that might be controlled.
RM: But if there isn't currently any means of controlling the
perception then the means doesn't exist (such a rocket that would take
people to the moon and back in 1960). And if the means does not exist
it doesn't exist in the environment. So it's not an effordance by your
definition which says it's "a condition in a person's environment".

MT: Even a
neutrino could qualify nowadays, perhaps not individually, but as a member
of a swarm, since neutrinos can now be detected and the result of their
detection is that some perceptions are altered, and it is possible to create
a setup so that a neutrino beam is aimed at a specific place. But neutrinos
were presumably just as abundant a century ago, and at that time they were
neither detectable nor manipulable, so I would guess that a century ago
neutrinos would provide an example of things in the environment that could
not be an effordance.
RM: Ignoring for the moment that a neutrino is a theoretical concept
and allowing the possibility that it is an actual environmental
entity, it was, by your definition, an effordance even a century ago,
since was (and is) a "condition in a person's environment through
which a person can control a perception".
If neutrinos were not
effordances when people didn't know how to use them to make neutrino
beams then Excel is not an effordance for people who don't know how to
use it to balance the books.

MT: There is probably a lot going on in our environment that has much the same
status now as neutrinos did a century ago.
RM: Right, and, more to the point, there are organizations of these
environmental variables that could help us control things better once
we figure out how to organize them properly; it's called invention.

 MT: You would want to bother [with effordances] if you wanted to help
someone control some variable (by saying something like "if you hold
the framingulus the other way up it might work better") or to make
controlling some variable more difficult (by taking the plank away
from the brook and destroying the footbridge).
RM: This just sounds like suggesting variables to control as the means
of controlling other variables. I think any student of PCT would
understand that suggesting lower level perceptions to control as the
means of controlling higher level ones would be a fundamental
component of teaching people to control. I don't see what effordance
adds.

MT: But here we are
getting into the social area in which we might consider Kent's
hypotheses. Do you want to do that?
RM: By all means. I have been trying to figure out what the science
(sociology) of effordance would be about. So how about describing some
data and showing me how the concept of effordance can contribute to
our understanding of what's going on.

__

[From Rick Marken (2012.11.30.0830)]

Martin Taylor (2012.11.29.23.52)--

MT: You would want to bother [with effordances] if you wanted to help
someone control some variable (by saying something like "if you hold
the framingulus the other way up it might work better") or to make
controlling some variable more difficult (by taking the plank away
from the brook and destroying the footbridge).

RM: This just sounds like suggesting variables to control as the means
of controlling other variables. I think any student of PCT would
understand that suggesting lower level perceptions to control as the
means of controlling higher level ones would be a fundamental
component of teaching people to control. I don't see what effordance
adds.

MT:I think you are looking at apples and calling them oranges because they are
both reasonably round objects of similar size. It's disingenuous.

RM: That doesn't help me understand how what you described is not
already explained by PCT without resort to the notion of effordance.
My little description of teaching uses only concepts from PCT and
explains what you say without using the concept of effordance. Saying
that I am looking at apples and calling them oranges doesn't help me
understand what your "apples" (effordances) provide that my "oranges"
(PCT) lack.

MT: Data follows theory or may guide toward theory, but it's the theory that is
paramount.

RM: I don't think it's a matter of which is more important; they are
both equally important, as far as I'm concerned. And both necessary.
Theory is just religion without data and data is just "a bunch of
stuff that happened" without theory.

MT: Is it "data" to say that of all the prisoners legally jailed, not
one has had a holiday on a beach in Samoa during the jail term without
escaping custody illegally.

RM: The proportion of prisoners going to Samoa would be data (I
presume it is 0.0 as you do); the statement is not.

MT: That's 100% accuracy for a statement that
control is impossible if the requisite effordances are removed by the action
of other control systems (assuming that some of the prisoners would have
liked to have a holiday on a beach in Samoa). Is it "data"?

RM: But I can (again) explain the same data using only PCT. The
feedback link from the prisoners' present location to Samoa is rather
effectively cut (though I'm sure there are prisoners who have managed
to escape and get to Samoa; so while the proportion of prison's
vacationing in Samoa may be 0.0, the proportion who flee to Samoa may
be somewhat larger, like .0001; but we would need the data) so they
are unable to control for that perception using the means that are
available to other travelers. The situation is exactly analogous to
what would happen in a tracking task if the connection from mouse to
cursor were suddenly cut; now you have no control over the
cursor/target relationship. If it really mattered to you to control
that relationship you might do some reorganizing -- substantial
reorganizing -- learn to program, write a new version of the program,
whatever -- and start controlling the cursor again.

The answer my friend is blowin' in the tracking task! We don't need
no stinkin' effordances;-)

MT: I find it odd to talk about a "science of effordance". Effordance is a
linking concept between a (presumed to exist) real world and the abstract
world of feedback functions. It isn't a science or even a theory.

RM: It's not even a attractive word;-)

MT: You may be
able to abstract some parameters of an effordance and use them to deduce the
form of a feedback function if you are lucky. But it's more probable that
you will be able to make predictions, such as that a person is likely to
cross the brook more slowly over a slippery mossy log than over a sturdy
solidly placed plank.

RM: But those predictions would also be made -- with quantitative
accuracy -- by PCT. Why introduce this confusing concept?

MT: I'm not clear how you would put that into a feedback
function other than to say that its transport lag is greater if the log
provides the effordance than if the effordance is provided by the plank.

RM: You would put it into the feedback function in terms of frictional
forces (far lower on the slippery log than the plank) and other
physical forces that would make walking over a slippery log a whole
different control task than walking over a dry flat plank. These
properties of the feedback connection all have to be part of the
control model of behavior, which they are in PCT.

MT: I think you tend to ask the wrong questions, questions based around the
ideas that effordances reside in objects as properties that can be measured
in the object. They don't. If a person has not the skill to use an object to
influence a perception, that object does not provide an effordance for that
person to control that perception. For controlling that perception, the
object might as well not exist. You simply can't talk about Excel providing
an effordance for balancing the books without at the same time saying "for a
person who wants to balance books and has the skill and the computational
support to use Excel."

RM: I think PCT does a fine job of explaining all the examples you
have brought up, including why a person who doesn't know Excel can't
use it to balance the book while a person who does know it, can. With
no need for the concept of effordance.

Apparently this concept of effordance is very important to you so I'm
sure you won't abandon it based on anything I (or anyone) says. So go
ahead and enjoy it; maybe it will make you famous. But it's of no
interest to me.

Best

Rick

···

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

[From Bill Powers (2012.11.30.1625 MST)]

Martin Taylor 2012.11.29.23.52 –

No, no. It is NOT, repeat NOT,
the same as the feedback function. A feedback function is a relationship
between an output variable and an input variable. It can be written i =
f(o, t). An effordance has characteristics that affect the form of the
feedback function, but is not the function.

I think the problem here is that we (I) have been too sloppy in defining
control systems in which the controlled variable is a function of
multiple lower-order variables. Maybe that’s because it’s not easy to
do.

You have said that an effordance has characteristics that affect the form
of the feedback function. Do you mean this?:

i <------Ff <---------o

\

← effordance

This shows an effordance as something separate from the feedback
function, affecting the function – presumably, altering some of
coefficients in the mathematical expression representing the physical
function. This makes the effordance not into something that intervenes
between o and i, but something that alters the function that actually
intervenes. Is this what you intend? If not, what do you mean by
“characteristics that affect the form”?

Another problem: the “input variable” i is being used as if it
is the controlled variable. Actually, there are multiple input variables
for any input function, no one of which is the controlled variable. The
output function generates an output that is fanned out through
various weights to effect each input variable. No one path connecting a
weighted output to one of the input variables is the feedback function.
In fact, the physical controlled variable is the perceptual signal, and
is not directly observable from outside the control system except by an
observer who employs the same Fi to generate his own perception from the
same set of input variables.

I’m getting swamped with email and other projects. If I skip some of
these interchanges it’s only because I am running out of available
time!

Tomorrow morning I get my first look at a completed apartment of the kind
I have been contemplating moving into. There is no guarantee that I will
go ahead with that, but if I do I’ll be even more swamped with trying to
sell the place I’m in now.

Best,

Bill P.