Analyzing feedback paths

Hi Fred,

what a nice surprise...Tahnks for your answers :slight_smile:

[Fred Nickols] I don't. I think the fielder/catcher is predicting whether
or not he can intercept the ball. He's not focused on the physical landing
point but, rather, the relative change in position between himself and the
ball.

Fred Nickols] Close but not quite. I don't think the catcher is estimating

the catching point. He's keeping track of how well he's reducing the
distance between himself and the ball, not where the ball will likely come
down.

[Fred Nickols] I don't. I think the fielder/catcher is predicting whether

or not he can intercept the ball. He's not focused on the physical landing
point but, rather, the relative change in position between himself and the
ball.

The initial trajectory

tells me something too; namely, the general direction, movement and speed.
Depending on what that is, I take off running, all the while keeping an eye
on the ball. The relative movement between me (one the run) and the ball
(on the fly) tell me if I'm likely to intercept it or not. If so, I keep
doing what I'm doing. If not, I speed up, change course, slow down or
whatever it takes to maintain my sense that I am likely to intercept the
ball. I do not estimate where it will land; I estimate or take stock of the
likelihood that I can/will intercept it.

HB:
Nice descriptions of the baseball game. Could you now explain everything
what you wrote with GAP_ACT model orconcept of whatever... or PCT model

Best,

Boris

路路路

----- Original Message ----- From: "Fred Nickols" <fred@NICKOLS.US>
To: <CSGNET@LISTSERV.ILLINOIS.EDU>
Sent: Tuesday, November 20, 2012 3:49 PM
Subject: Re: Analyzing feedback paths

[From Fred Nickols (2012.11.20.0733 AZ)]

My comments are embedded below. Material to which I am not responding has
been snipped.

-----Original Message-----
From: Control Systems Group Network (CSGnet)
[mailto:CSGNET@LISTSERV.ILLINOIS.EDU] On Behalf Of boris_upc
Sent: Tuesday, November 20, 2012 7:25 AM
To: CSGNET@LISTSERV.ILLINOIS.EDU
Subject: Re: Analyzing feedback paths

O.K. I think that we need unique theory of ball-catching (human behavior)

[Fred Nickols] I think we have a theory of ball-catching, as expressed in
Rick Marken's model.

"OLD" THEORY :
BP :

2. I have satisfied myself by watching baseball on television that
> catchers (more properly, "fielders") who try to catch balls in the way
> the old theory assumes usually fail to catch it. Sometimes they
> succeed, and such successes are often remembered exactly because they
are so rare.

[Fred Nickols] Actually, I don't think Bill can say what he wrote above.
As he likes to point out, we can't tell what someone is up to by simply
observing their behavior. He might "guess" that some fielders are trying to
estimate where the ball will come down but I don't think he can really tell
that's what they're trying to do.

HB:
Both theories (if I understand right) explain special cases or maybe I can

say

special "playing situations". And as you say they are experienced (on TV),

I

can probably conclude that both are right, as they correctly predict or
describe behavior of "real" players in "real" situation.
I suppose that only the theory which describes or predicts behavior in the
way that matches the "reality" is for me right theory.

In 1.case you say (if I understand right) that ball (*ffordance) is

"directing"

behavior of the catcher (fielder), because the catcher (fielder) is

"looking" at

the ball all the time (with some constant or slowly changing
angle) and fielder "miraculously" find himself "exactly where the ball

comes

down or close enough to catch it". I somehow missed PCT explanation of

this

"situation".

[Fred Nickols] Again, I think that's Rick's model.

I think that in both cases catcher (fielder) is estimating the

"catching-point" as

the goal of every behavior of the fielder in any case is to "catch the

ball".

[Fred Nickols] I'm not a good ball player but I have played some and in the
outfield. The first thing to which I recall responding is the crack of the
bat hitting the ball. The sound tells me something. The initial trajectory
tells me something too; namely, the general direction, movement and speed.
Depending on what that is, I take off running, all the while keeping an eye
on the ball. The relative movement between me (one the run) and the ball
(on the fly) tell me if I'm likely to intercept it or not. If so, I keep
doing what I'm doing. If not, I speed up, change course, slow down or
whatever it takes to maintain my sense that I am likely to intercept the
ball. I do not estimate where it will land; I estimate or take stock of the
likelihood that I can/will intercept it.

We don't need two theories.

[Fred Nickols]
I don't think Bill was advancing two theories but I'll leave that to him.

I'll try with Fred's GAP-ACT concept to make a unigue concept (model),

which

would be able to explain every behavior of the catchers (fielders). I'll

try in

this way :

INSIDE ORGANISM
G = goal (wanted perception, catching the ball or the ball in the glove of

the

catcher) A = action (internal environment. I suppose that Fred ment, if I

see

right, nervous-muscles connection, driven by "error" in comparator) P =
perception (I suppose perception of the ball)

OUTSIDE ORGANISM
A = action (behavior, outer expression of the muscles activity in outer
environment that can be observed) C = conditions (probably Fred meant
variables in environment that affect the ball - disturbances - and other
environmental variables) T = target (I suppose he meant controlled

variable

in outer environment, the
ball)

[Fred Nickols] Fred meant "ball in glove" or, in shorthand, "caught"

So my opppinion is that the catcher (fielder) is perceiving the flight of

the

ball, and in imagination starts to estimate the place where the ball will

fall.

[Fred Nickols] I don't. I think the fielder/catcher is predicting whether
or not he can intercept the ball. He's not focused on the physical landing
point but, rather, the relative change in position between himself and the
ball.

There are probably differences in how fast "players" estimates the

"catching

point".

When the goal is set (wanted perception of the ball), the difference
between goal and actual flight of the ball is perceived, so with action

the

fielder starts to reduce the difference between actual perception of the
flying ball and reference - estimated catching point in imagination.

[Fred Nickols] Close but not quite. I don't think the catcher is estimating
the catching point. He's keeping track of how well he's reducing the
distance between himself and the ball, not where the ball will likely come
down.

If in your 1. case, "looking at the ball" (for ex. it's "angular

velocity") is the

reference of the fielder, from what will perception be subtracted ?

So as I see it, every behavior by Fred model (concept) can be explained,

so

the theory should be working for all goal-directed behaviors of human.

Can you use the same Freds' concept to explain your vision what is
happening. You can make also mathematical description of what is
happening.
I'm really interested to see it.

[Fred Nickols] Boris: I'm glad you like my GAP-ACT model but I don't think
it's quite fair to ask Bill to explain anything in terms of it. Besides,
it's not "my" concept; it's simply a view based on PCT.

[From Fred Nickols (2012.11.20.0836 AZ)]

I'll try, Boris.

In terms of the GAP-ACT Model, the target variable (T) is the distance
between the fielder and the ball. The goal (G) is for that distance to be
sufficiently small so as to enable catching the ball. The actions (A)
consist of running, altering direction and speed, keeping an eye on the
ball, monitoring the distance reduction, etc. The perceptions of interest
(P) are the distance between the fielder and the ball and the rate of
closing any gap. If the distance is increasing despite the fielder's best
efforts, the ball is not going to be caught, at which point the fielder
might do something like predict where the ball will hit and how it might
bounce (e.g., off the outfield wall or into it) and he will head for that
point. The conditions (C) might include the wind (causing the ball to drift
but compensated for by the running), and the surface of the outfield (e.g.,
the ball takes a weird bounce and I think most of us have seen outfielders
surprised by that "disturbance").

Again, please keep in mind, Boris, that what I am offering is not a model of
outfielder behavior; I'm simply examining what I can observe and explaining
it in terms of the GAP-ACT (PCT) framework. My explanations could be right
on the money or woefully misbegotten.

Fred Nickols

From: Control Systems Group Network (CSGnet)
[mailto:CSGNET@LISTSERV.ILLINOIS.EDU] On Behalf Of boris_upc
Sent: Tuesday, November 20, 2012 8:31 AM
To: CSGNET@LISTSERV.ILLINOIS.EDU
Subject: Re: Analyzing feedback paths

Hi Fred,

what a nice surprise...Tahnks for your answers :slight_smile:

[Fred Nickols] I don't. I think the fielder/catcher is predicting

whether or not

he can intercept the ball. He's not focused on the physical landing point

but,

rather, the relative change in position between himself and the ball.

Fred Nickols] Close but not quite. I don't think the catcher is

estimating

> the catching point. He's keeping track of how well he's reducing the
> distance between himself and the ball, not where the ball will likely
> come down.

[Fred Nickols] I don't. I think the fielder/catcher is predicting whether
> or not he can intercept the ball. He's not focused on the physical
> landing point but, rather, the relative change in position between
> himself and the ball.

The initial trajectory
> tells me something too; namely, the general direction, movement and
speed.
> Depending on what that is, I take off running, all the while keeping
> an eye on the ball. The relative movement between me (one the run)
> and the ball (on the fly) tell me if I'm likely to intercept it or
> not. If so, I keep doing what I'm doing. If not, I speed up, change
> course, slow down or whatever it takes to maintain my sense that I am
> likely to intercept the ball. I do not estimate where it will land; I
> estimate or take stock of the likelihood that I can/will intercept it.

HB:
Nice descriptions of the baseball game. Could you now explain everything
what you wrote with GAP_ACT model orconcept of whatever... or PCT model

Best,

Boris

From: "Fred Nickols" <fred@NICKOLS.US>
To: <CSGNET@LISTSERV.ILLINOIS.EDU>
Sent: Tuesday, November 20, 2012 3:49 PM
Subject: Re: Analyzing feedback paths

> [From Fred Nickols (2012.11.20.0733 AZ)]
>
> My comments are embedded below. Material to which I am not
responding has
> been snipped.
>
>> From: Control Systems Group Network (CSGnet)
>> [mailto:CSGNET@LISTSERV.ILLINOIS.EDU] On Behalf Of boris_upc
>> Sent: Tuesday, November 20, 2012 7:25 AM
>> To: CSGNET@LISTSERV.ILLINOIS.EDU
>> Subject: Re: Analyzing feedback paths
>>
>> O.K. I think that we need unique theory of ball-catching (human

behavior)

>
> [Fred Nickols] I think we have a theory of ball-catching, as expressed

in

> Rick Marken's model.
>>
>> "OLD" THEORY :
>> BP :
>>
>> 2. I have satisfied myself by watching baseball on television that
>> > catchers (more properly, "fielders") who try to catch balls in the

way

>> > the old theory assumes usually fail to catch it. Sometimes they
>> > succeed, and such successes are often remembered exactly because
they
>> are so rare.
>
> [Fred Nickols] Actually, I don't think Bill can say what he wrote

above.

> As he likes to point out, we can't tell what someone is up to by simply
> observing their behavior. He might "guess" that some fielders are

trying

> to
> estimate where the ball will come down but I don't think he can really
> tell
> that's what they're trying to do.
>
>>
>> HB:
>> Both theories (if I understand right) explain special cases or maybe I
>> can
> say
>> special "playing situations". And as you say they are experienced (on
>> TV),
> I
>> can probably conclude that both are right, as they correctly predict or
>> describe behavior of "real" players in "real" situation.
>> I suppose that only the theory which describes or predicts behavior in
>> the
>> way that matches the "reality" is for me right theory.
>>
>> In 1.case you say (if I understand right) that ball (*ffordance) is
> "directing"
>> behavior of the catcher (fielder), because the catcher (fielder) is
> "looking" at
>> the ball all the time (with some constant or slowly changing
>> angle) and fielder "miraculously" find himself "exactly where the ball
> comes
>> down or close enough to catch it". I somehow missed PCT explanation of
> this
>> "situation".
>
> [Fred Nickols] Again, I think that's Rick's model.
>>
>> I think that in both cases catcher (fielder) is estimating the
> "catching-point" as
>> the goal of every behavior of the fielder in any case is to "catch the
> ball".
>
> [Fred Nickols] I'm not a good ball player but I have played some and in
> the
> outfield. The first thing to which I recall responding is the crack of
> the
> bat hitting the ball. The sound tells me something. The initial
> trajectory
> tells me something too; namely, the general direction, movement and
speed.
> Depending on what that is, I take off running, all the while keeping an
> eye
> on the ball. The relative movement between me (one the run) and the

ball

> (on the fly) tell me if I'm likely to intercept it or not. If so, I

keep

> doing what I'm doing. If not, I speed up, change course, slow down or
> whatever it takes to maintain my sense that I am likely to intercept the
> ball. I do not estimate where it will land; I estimate or take stock of
> the
> likelihood that I can/will intercept it.
>
>> We don't need two theories.
> [Fred Nickols]
> I don't think Bill was advancing two theories but I'll leave that to

him.

>>
>> I'll try with Fred's GAP-ACT concept to make a unigue concept (model),
> which
>> would be able to explain every behavior of the catchers (fielders).

I'll

> try in
>> this way :
>>
>> INSIDE ORGANISM
>> G = goal (wanted perception, catching the ball or the ball in the glove
>> of
> the
>> catcher) A = action (internal environment. I suppose that Fred ment, if

I

> see
>> right, nervous-muscles connection, driven by "error" in comparator) P =
>> perception (I suppose perception of the ball)
>>
>> OUTSIDE ORGANISM
>> A = action (behavior, outer expression of the muscles activity in outer
>> environment that can be observed) C = conditions (probably Fred meant
>> variables in environment that affect the ball - disturbances - and

other

>> environmental variables) T = target (I suppose he meant controlled
> variable
>> in outer environment, the
>> ball)
>
> [Fred Nickols] Fred meant "ball in glove" or, in shorthand, "caught"
>>
>> So my opppinion is that the catcher (fielder) is perceiving the flight

of

> the
>> ball, and in imagination starts to estimate the place where the ball

will

> fall.
>
> [Fred Nickols] I don't. I think the fielder/catcher is predicting

whether

> or not he can intercept the ball. He's not focused on the physical
> landing
> point but, rather, the relative change in position between himself and

the

> ball.
>
>> There are probably differences in how fast "players" estimates the
> "catching
>> point".
>>
>> When the goal is set (wanted perception of the ball), the difference
>> between goal and actual flight of the ball is perceived, so with action
> the
>> fielder starts to reduce the difference between actual perception of

the

>> flying ball and reference - estimated catching point in imagination.
>
> [Fred Nickols] Close but not quite. I don't think the catcher is
> estimating
> the catching point. He's keeping track of how well he's reducing the
> distance between himself and the ball, not where the ball will likely

come

> down.
>
>>
>> If in your 1. case, "looking at the ball" (for ex. it's "angular
> velocity") is the
>> reference of the fielder, from what will perception be subtracted ?
>>
>> So as I see it, every behavior by Fred model (concept) can be

explained,

> so
>> the theory should be working for all goal-directed behaviors of human.
>>
>> Can you use the same Freds' concept to explain your vision what is
>> happening. You can make also mathematical description of what is
>> happening.
>> I'm really interested to see it.
>>
> [Fred Nickols] Boris: I'm glad you like my GAP-ACT model but I don't
> think
> it's quite fair to ask Bill to explain anything in terms of it.

Besides,

路路路

-----Original Message-----
----- Original Message -----
>> -----Original Message-----
> it's not "my" concept; it's simply a view based on PCT.
>

Thanks Fred for your effort. I think it's nice presentation
of what is happening in the baseball game. I don't only
understand why you are under-estimating your explanation
and GAP-ACT concept. I like it.

As I'm not a player nor the fan of the baseball, I'll wait for
the oppinion of others. Maybe there are some who
played or are activelly playing baseball. Their experiences
could be precious for the understanding of the game...:wink:

Best,

Boris

路路路

----- Original Message ----- From: "Fred Nickols" <fred@NICKOLS.US>
To: <CSGNET@LISTSERV.ILLINOIS.EDU>
Sent: Tuesday, November 20, 2012 4:46 PM
Subject: Re: Analyzing feedback paths

[From Fred Nickols (2012.11.20.0836 AZ)]

I'll try, Boris.

In terms of the GAP-ACT Model, the target variable (T) is the distance
between the fielder and the ball. The goal (G) is for that distance to be
sufficiently small so as to enable catching the ball. The actions (A)
consist of running, altering direction and speed, keeping an eye on the
ball, monitoring the distance reduction, etc. The perceptions of interest
(P) are the distance between the fielder and the ball and the rate of
closing any gap. If the distance is increasing despite the fielder's best
efforts, the ball is not going to be caught, at which point the fielder
might do something like predict where the ball will hit and how it might
bounce (e.g., off the outfield wall or into it) and he will head for that
point. The conditions (C) might include the wind (causing the ball to drift
but compensated for by the running), and the surface of the outfield (e.g.,
the ball takes a weird bounce and I think most of us have seen outfielders
surprised by that "disturbance").

Again, please keep in mind, Boris, that what I am offering is not a model of
outfielder behavior; I'm simply examining what I can observe and explaining
it in terms of the GAP-ACT (PCT) framework. My explanations could be right
on the money or woefully misbegotten.

Fred Nickols

-----Original Message-----
From: Control Systems Group Network (CSGnet)
[mailto:CSGNET@LISTSERV.ILLINOIS.EDU] On Behalf Of boris_upc
Sent: Tuesday, November 20, 2012 8:31 AM
To: CSGNET@LISTSERV.ILLINOIS.EDU
Subject: Re: Analyzing feedback paths

Hi Fred,

what a nice surprise...Tahnks for your answers :slight_smile:

[Fred Nickols] I don't. I think the fielder/catcher is predicting

whether or not

he can intercept the ball. He's not focused on the physical landing point

but,

rather, the relative change in position between himself and the ball.

Fred Nickols] Close but not quite. I don't think the catcher is

estimating

> the catching point. He's keeping track of how well he's reducing the
> distance between himself and the ball, not where the ball will likely
> come down.

[Fred Nickols] I don't. I think the fielder/catcher is predicting whether
> or not he can intercept the ball. He's not focused on the physical
> landing point but, rather, the relative change in position between
> himself and the ball.

The initial trajectory
> tells me something too; namely, the general direction, movement and
speed.
> Depending on what that is, I take off running, all the while keeping
> an eye on the ball. The relative movement between me (one the run)
> and the ball (on the fly) tell me if I'm likely to intercept it or
> not. If so, I keep doing what I'm doing. If not, I speed up, change
> course, slow down or whatever it takes to maintain my sense that I am
> likely to intercept the ball. I do not estimate where it will land; I
> estimate or take stock of the likelihood that I can/will intercept it.

HB:
Nice descriptions of the baseball game. Could you now explain everything
what you wrote with GAP_ACT model orconcept of whatever... or PCT model

Best,

Boris

----- Original Message -----
From: "Fred Nickols" <fred@NICKOLS.US>
To: <CSGNET@LISTSERV.ILLINOIS.EDU>
Sent: Tuesday, November 20, 2012 3:49 PM
Subject: Re: Analyzing feedback paths

> [From Fred Nickols (2012.11.20.0733 AZ)]
>
> My comments are embedded below. Material to which I am not
responding has
> been snipped.
>
>> -----Original Message-----
>> From: Control Systems Group Network (CSGnet)
>> [mailto:CSGNET@LISTSERV.ILLINOIS.EDU] On Behalf Of boris_upc
>> Sent: Tuesday, November 20, 2012 7:25 AM
>> To: CSGNET@LISTSERV.ILLINOIS.EDU
>> Subject: Re: Analyzing feedback paths
>>
>> O.K. I think that we need unique theory of ball-catching (human

behavior)

>
> [Fred Nickols] I think we have a theory of ball-catching, as expressed

in

> Rick Marken's model.
>>
>> "OLD" THEORY :
>> BP :
>>
>> 2. I have satisfied myself by watching baseball on television that
>> > catchers (more properly, "fielders") who try to catch balls in the

way

>> > the old theory assumes usually fail to catch it. Sometimes they
>> > succeed, and such successes are often remembered exactly because
they
>> are so rare.
>
> [Fred Nickols] Actually, I don't think Bill can say what he wrote

above.

> As he likes to point out, we can't tell what someone is up to by simply
> observing their behavior. He might "guess" that some fielders are

trying

> to
> estimate where the ball will come down but I don't think he can really
> tell
> that's what they're trying to do.
>
>>
>> HB:
>> Both theories (if I understand right) explain special cases or maybe I
>> can
> say
>> special "playing situations". And as you say they are experienced (on
>> TV),
> I
>> can probably conclude that both are right, as they correctly predict >> or
>> describe behavior of "real" players in "real" situation.
>> I suppose that only the theory which describes or predicts behavior in
>> the
>> way that matches the "reality" is for me right theory.
>>
>> In 1.case you say (if I understand right) that ball (*ffordance) is
> "directing"
>> behavior of the catcher (fielder), because the catcher (fielder) is
> "looking" at
>> the ball all the time (with some constant or slowly changing
>> angle) and fielder "miraculously" find himself "exactly where the ball
> comes
>> down or close enough to catch it". I somehow missed PCT explanation of
> this
>> "situation".
>
> [Fred Nickols] Again, I think that's Rick's model.
>>
>> I think that in both cases catcher (fielder) is estimating the
> "catching-point" as
>> the goal of every behavior of the fielder in any case is to "catch >> the
> ball".
>
> [Fred Nickols] I'm not a good ball player but I have played some and in
> the
> outfield. The first thing to which I recall responding is the crack of
> the
> bat hitting the ball. The sound tells me something. The initial
> trajectory
> tells me something too; namely, the general direction, movement and
speed.
> Depending on what that is, I take off running, all the while keeping an
> eye
> on the ball. The relative movement between me (one the run) and the

ball

> (on the fly) tell me if I'm likely to intercept it or not. If so, I

keep

> doing what I'm doing. If not, I speed up, change course, slow down or
> whatever it takes to maintain my sense that I am likely to intercept > the
> ball. I do not estimate where it will land; I estimate or take stock > of
> the
> likelihood that I can/will intercept it.
>
>> We don't need two theories.
> [Fred Nickols]
> I don't think Bill was advancing two theories but I'll leave that to

him.

>>
>> I'll try with Fred's GAP-ACT concept to make a unigue concept (model),
> which
>> would be able to explain every behavior of the catchers (fielders).

I'll

> try in
>> this way :
>>
>> INSIDE ORGANISM
>> G = goal (wanted perception, catching the ball or the ball in the >> glove
>> of
> the
>> catcher) A = action (internal environment. I suppose that Fred ment, >> if

I

> see
>> right, nervous-muscles connection, driven by "error" in comparator) P >> =
>> perception (I suppose perception of the ball)
>>
>> OUTSIDE ORGANISM
>> A = action (behavior, outer expression of the muscles activity in >> outer
>> environment that can be observed) C = conditions (probably Fred meant
>> variables in environment that affect the ball - disturbances - and

other

>> environmental variables) T = target (I suppose he meant controlled
> variable
>> in outer environment, the
>> ball)
>
> [Fred Nickols] Fred meant "ball in glove" or, in shorthand, "caught"
>>
>> So my opppinion is that the catcher (fielder) is perceiving the flight

of

> the
>> ball, and in imagination starts to estimate the place where the ball

will

> fall.
>
> [Fred Nickols] I don't. I think the fielder/catcher is predicting

whether

> or not he can intercept the ball. He's not focused on the physical
> landing
> point but, rather, the relative change in position between himself and

the

> ball.
>
>> There are probably differences in how fast "players" estimates the
> "catching
>> point".
>>
>> When the goal is set (wanted perception of the ball), the difference
>> between goal and actual flight of the ball is perceived, so with >> action
> the
>> fielder starts to reduce the difference between actual perception of

the

>> flying ball and reference - estimated catching point in imagination.
>
> [Fred Nickols] Close but not quite. I don't think the catcher is
> estimating
> the catching point. He's keeping track of how well he's reducing the
> distance between himself and the ball, not where the ball will likely

come

> down.
>
>>
>> If in your 1. case, "looking at the ball" (for ex. it's "angular
> velocity") is the
>> reference of the fielder, from what will perception be subtracted ?
>>
>> So as I see it, every behavior by Fred model (concept) can be

explained,

> so
>> the theory should be working for all goal-directed behaviors of human.
>>
>> Can you use the same Freds' concept to explain your vision what is
>> happening. You can make also mathematical description of what is
>> happening.
>> I'm really interested to see it.
>>
> [Fred Nickols] Boris: I'm glad you like my GAP-ACT model but I don't
> think
> it's quite fair to ask Bill to explain anything in terms of it.

Besides,

> it's not "my" concept; it's simply a view based on PCT.
>

[From Bill Powers (2012.11.20.1604 MST)]

HB: O.K. I think that we need unique theory of ball-catching (human behavior)

As far as I can see you are proposing two different theories:

"NEW" THEORY :
BP:

1. It is not necessary for the

catcher to estimate where the ball will come down. If the catcher moves so
as to keep the apparent position of the ball vertically above home plate
at some constant or slowly changing angle, the catcher will find himself
exactly where the ball comes down, or close enough to catch it.

"OLD" THEORY :
BP :

2. I have satisfied myself by watching baseball on television that

catchers (more properly, "fielders") who try to catch balls in the way the
old theory assumes usually fail to catch it. Sometimes they succeed, and
such successes are often remembered exactly because they are so rare.

BP: The "old" theory is the one most scientists have believed and which I think is wrong. Some baseball players believe it, and when they try to catch a ball that way (by running to the place they predict it will come down) they usually can't catch it. The right way, and the one that just about always works to catch a ball that is possible to catch, is the first one. The player adjusts the direction and speed of running first so the ball appears to be rising straight up (left-right angle remains constant relative to objects on the ground), and then so it rises only very slowly or stays at the same vertical visual angle relative to objects on the ground.

Those are the two perceptions that have to be controlled by varying speed and direction of running. It is easy to do and requires no predictions or calculations about the path of the ball or where it will come down. Players who consistently catch fly balls probably have organized themselves the "New" way.

HB:
Both theories (if I understand right) explain special cases or maybe I can
say special "playing situations". And as you say they are experienced (on
TV), I can probably conclude that both are right, as they correctly predict
or describe behavior of "real" players in "real" situation.
I suppose that only the theory which describes or predicts behavior in the way
that matches the "reality" is for me right theory.

BP: But one of them doesn't work well, and does not normally result in catching the ball. Very few successful fielders use it. This is because it's impossible for a fielder to estimate accurately enough where a fly ball is going to land. If the fielder does catch the ball is it only because the random scatter of estimated landing points happens to include the actual landing point and the fielder is lucky once in a while.

I recommend that you go outdoors with a ball and a friend and try this out. If you have trouble catching balls and are trying to do it by predicting the path of the ball, you will find that the constant-direction works like magic, bringing you to the right place every time the ball can actually be caught.

HB: In 1.case you say (if I understand right) that ball (*ffordance) is
"directing" behavior of the catcher (fielder), because the catcher (fielder)
is "looking" at the ball all the time (with some constant or slowly changing
angle) and fielder "miraculously" find himself "exactly where the ball comes
down or close enough to catch it". I somehow missed PCT explanation of this
"situation".

In case 1, the reference condition for one control system specifies that the ball is to be stationary in the left-right direction. If the perceived velocity is to the right, the negative error is converted into a reference signal at a lower level for turning the direction of running toward the left. A positive error results in changing direction to the right. When the perceived direction equals the reference direction (zero left-right velocity), the direction of the ball's trajectory is vertical, either rising or falling. A control system with an integrating output function is needed to do this.

For control in the up-down direction, a similar design is needed, with the reference signal indicating an angle above the playing field, and the error signal causing running toward the ball or away from the ball depending on the sign of the error. The first control system keeps the fielder in the (vertical) plane of the ball's trajectory, while the second one then changes the running speed over a range between positive and negative limits.

This sort of arrangement will in fact result in the fielder's following a path that places him at the exact spot where the ball is arriving. No magic or miracles are required.

I think you can see that if the ball is not moving either to the left or the right, the fielder must be in the plane of the ball's arching trajectory. When the ball lands, it will still be in that plane, but the fielder might be too close or too far away.

The second control system keeps the apparent rate of change of the vertical angle close to zero. If you draw a diagram showing the ball's trajectory, with the player standing at any distance, you will see that after the ball passes its maximum altitude and starts down again, the direction of the vertical angle between the fielder's head and the ball depends on whether the fielder is too close or too far away. If the fielder is too far away, that angle will be decreasing, so the error signal in that system should make the fielder run closer, and the opposite if the fielder is too close. So both control systems are quite simple, and when they both experience zero error when the ball comes down, the ball will hit the fielder in his head, or in the dominant eye.

HB: In 2. case (if I understand right) you are saying that you saw on TV how
catchers (fielders) are estimating the flight of the ball and run to the place where
they can catch it. And these succesfull catching are very rare.

My oppinion is, that even if behavior is rare, it's part of the control
system which is functioning in the same way in any case (1. and 2. case) and
any behavior control system produces that we can perceive or imagine is
valid and must be included in theory and explained with the same "tools", if
the theory is right and good.

BP: It is rare because it works so poorly; it seldom results in actually catching the ball. So if a control system is organized to work that way, it will experience many large errors, and will be reorganized until it works the right way.

HB: If PCT is believed to be right and unique theory of human behavior, than I
suppose it has to explain any "real" behavior of control system in succesfull way.
Although I suppose it's better one PCT way for explaining all behaviors
not two different ways (for ex. old and new).
I don't see how can "real" control system work in separate cases differently
(in new and old way) if there is only one PCT and if "real situation" is reference
for the theory to be right or wrong.

BP: You are forgetting about reorganization. Control systems are not wired into the brain so they can work only one way. They are a result of contininuous reorganization that eventually finds a way to reduce errors to zero, or close enough. Only a few of the hierarchical control systems are working when a person is born, and even those have to be reorganized to work properly when interactions with the environment start.

HB: If you saw it on TV, than it doesn't matter how rare behavior is, it is
something what happened "in "reality" and control system did it, so if the
theory is right it will succesfully explain what has happened.
So I think that succesfull explanation with the unique theory which explain
ALL behaviors of control system, is right and good.

And PCT is supposed to be that kind of theory.

BP: It seems that you have been thinking that all these control systems exist in the brain from the beginning of life. I assume that only the potential for control systems is there at first -- the neural and muscular building blocks from which control systems can be created. The actual control systems we find in adults are, I assume, the result of long years of reorganization with both successes and failures happening again and again along the journey. Any organization is possible. Tom Bourbon once showed that a control system can be built that imitates a stimulus-response system. But stimulus-response systems can't seek goals and they can't produce repeatable results in a changeable environment or when unpredictable disturbances occur, as they always do. So that kind of organization will be reorganized away and replaced by a control system that can do those things.

I think that in both cases catcher (fielder) is estimating the
"catching-point" as the goal of every behavior of the fielder in any case is to
"catch the ball".

Wanting to catch the ball is far from knowing where the ball will come down. I strongly doubt that any successful fielder is organized that way.

In 1. case he is estimating the catching-point on the bases of "continuous
perception", as he has time to "watch the ball". The more he watch the ball
the more precise he will estimate the "catching-point", and the more chances
that trial of "catching the ball" will be succesfull.

OK, so you have chosen to believe the old theory, the one you will find in most textbooks on motor control. If you can produce a model that works that way and can handle things like winds that disturb the path of the ball, tell us about it. You could be right, even if I doubt it. I'd like to see what the necessary perceptions are, and the computations, and what accuracy and speed of computing are required. I know what is required for the other model, and it's pretty simple. Yours sounds immensely complex.

HB: In the 2. case as I see it, fielder hasn't enough time to precisly estimate
the flight of the ball, because it's probably to fast or/and to high. So he saw it for the moment and start running immediatelly as he estimated that any part of the
second that he miss, the controlling of the "catching a ball" will be
impossible. So he starts running maybe "blindly" or by some "feeling" to the
"catching-point".

But as I see it, the basis of "estimation of the flight of the ball" is the
same in all behaviors, no matter how much time fielder has to estimate it's
trajectory. The process of control that's going on in organism is the same
in any case, if PCT is right theory.

So in any case as I see it, fielder is controlling for "catching the ball"
in certain point. If you'll watch on TV again, I'm sure you will precisely
see that any player in any "playing situation" is controling for "catching the ball".

BP: Sure, but that alone isn't enough to make him able to catch it. The player is also controlling for "winning the game" even if his team has lost 20 games in a row. A professional fielder probably catches a high percentage of catchable balls, but the ony way to do that is by using the best way to do it.

HB: So I think that at least we must have some unigue model (concept) that will
explain all the behaviors of fielders if the theory is matching the
"reality" ? We don't need two theories.

I'll try with Fred's GAP-ACT concept to make a unigue concept (model), which would
be able to explain every behavior of the catchers (fielders). I'll try in
this way :
If in your 1. case, "looking at the ball" (for ex. it's "angular velocity") is the referenceof the fielder, from what will perception be subtracted ?

BP: If the reference signal is set to zero, the angle from which the perceived angle is subtracted is zero. This would be the case for the system that controls angular velocity in the horizontal direction. For the vertically-oriented angle, the reference signal would have a value corresponding to whatever angle has been found to work well enough.

So as I see it, every behavior by Fred model (concept) can be explained, so
the theory should be working for all goal-directed behaviors of human.

BP: But you haven't offered an actual explanation, have you? You've sketched in some of the things your model has to accomplish, but you haven't said anything that would lead to an actual working model.

HB: Can you use the same Freds' concept to explain your vision what is
happening. You can make also mathematical description of what is happening.
I'm really interested to see it.

Of course we have the mathematical description. How do you think those models that Rick and I work out end up actually working? The programs contain the mathematical operations that we proposed to be carried out by neural circuits. Rick has actual data from other researchers, measurements obtained from cameras carried by fielders as they run to catch balls, and other videos that show their paths and speeds across the playing field. Those give us the numbers that the model has to match by its own behavior. I'll let Rick show you the programs he is using, since I'm just watching from the sidelines.

HB: I'm most interested in PCT explanation of this "new" one :

BP: Actually this is the old theory of ball-catching that Rick's
model and other related ones refute. It is not necessary for the
catcher to estimate where the ball will come down. If the catcher
moves so as to keep the apparent position of the ball vertically
above home plate at some constant or slowly changing vertical angle, the
catcher will find himself exactly where the ball comes down, or close
enough to catch it.

HB :
What excatly is happening in the comparator while catcher (fielder) "moves
so as to keep the apparent position of the ball.".
What is so apparent that it don't need PCT explanation ?\

BP: You'll find most of it earlier in this post. In the comparator, what happens is that the perceptual signal is rapidly altered by the changes in running speed and direction, until the error gets close to zero.

The word "apparent" was used as opposed to "actual". As the fielder moves, the angle of the ball as the fielder sees it seems to change up, down, left, and right, but that is only the way it looks. Its actual path through space doesn't change that way. The apparent changes are caused by the changing relationship between the fielder and the ball.

Best,

Bill P.

路路路

At 03:24 PM 11/20/2012 +0100, boris_upc wrote:

[From Bill Powers (2012.11.20.1604 MST)]

HB: O.K. I think that we need unique theory of ball-catching (human
behavior) As far as I can see you are proposing two different theories:

"NEW" THEORY :
BP:
1. It is not necessary for the

catcher to estimate where the ball will come down. If the catcher moves so
as to keep the apparent position of the ball vertically above home plate
at some constant or slowly changing angle, the catcher will find himself
exactly where the ball comes down, or close enough to catch it.

"OLD" THEORY :
BP :
2. I have satisfied myself by watching baseball on television that

catchers (more properly, "fielders") who try to catch balls in the way the
old theory assumes usually fail to catch it. Sometimes they succeed, and
such successes are often remembered exactly because they are so rare.

BP: The "old" theory is the one most scientists have believed and
which I think is wrong.

HB :
You think it's wrong ? It seems to me, that you are pretty unsure about that
? Who are those "most" scientist ?
Is the "old theory" - "running to the place they predict the ball will fall
down" ? Do I understand right that the "old theory" has for goal "catching
the ball" or "intercepting it" ?

I thought this is a PCT theory where reference is something "presumably
determined in your brain" (LCS III, p. 21). So predicting the place where
the ball will fall down or "the catching point" or "interecepting point" is
something "presumably determined in our brain".

I like your "NEW" definition of reference signal :
"The intended value of the perceptual signal is indicated by another
signal of exactly the same kind as the perceptual signal, but with a
different source. This other signal, the reference signal, is an example of
the perceptual signal's state as it would be when the controlled variable is
in its intended state. (LCS III, p.31).

This seems to me perfect description of goal, which could be "catching" or
"intercetping" the ball as some "intended" state (catching
point) is "construction" in the brain...

BP :
Some baseball players believe it, and when they try to catch a ball that way
(by running to the place they predict it will come down) they usually can't
catch it.

HB :
Well this one will be hard to prove. Do you really think that some
rare behaviors are "attached" to specific baseball players ?
Do you maybe know who are those some baseball players
that "believe in rare behaviors" ?
Maybe one name ? Can Rick give some names as he has so much
material upon baseball catch as you wrote in the end of your post ?

Rare behaviors by my oppinion take its source in "rare situation" which
consists of players abilities, his "better or worse position" in the
relation to the characteristics of ball that is hit.
I believe that every player in such "rare position" would exhibit
reorganized "rare behavior" as his goal is to "catch the ball" not to
exhibit "rare behavior".

I have a feeling that you are trying to "squeeze" nature into the "NEW"
theory (compare to B:CP, 2005, p.11).

BP : The right way, and the one that just about always works to catch a ball
that is possible to catch, is the first one..//.Those are the two
perceptions that have to be controlled by varying speed and direction of
running.
It is easy to do and requires no predictions or calculations about the path
of the ball or where it will come down. Players who consistently catch fly
balls
probably have organized themselves the "New" way.

HB :
Does "two perceptions that have to be controlled" means that players are
observing the ball all the time and ball is determining players behavior in
every moment ?

What does it mean "PROBABLY" have organized themselves the "NEW way" ?

BP: But one of them doesn't work well, and does not normally result
in catching the ball. Very few successful fielders use it. This is
because it's impossible for a fielder to estimate accurately enough
where a fly ball is going to land. If the fielder does catch the ball
is it only because the random scatter of estimated landing points
happens to include the actual landing point and the fielder is lucky
once in a while.

HB:
Does this mean that behaviors which are resulting in "rare and few sucessful
trials" should be excluded from "NEW" theory ? Can "NEW" theory explain
those
"rare" behaviors, speccially those when fielder is not observing the ball ?

BP :
I recommend that you go outdoors with a ball and a friend and try
this out. If you have trouble catching balls and are trying to do it
by predicting the path of the ball, you will find that the
constant-direction works like magic, bringing you to the right place
every time the ball can actually be caught.

HB :
What a good idea. I tried a little with my students and I have to improvize.
The ball was not original. It was tennis ball, and the bat was some wood
stick. We didn't play on the real baseball ground. It was on the meadow
with short grass and we were only hitting to catch the ball.
Although throws were easy, we have much troubles with
getting used to hitting and catching the ball. So we'll try to repeat it.
But one thing is sure. If ball was hit toward someone,
and it was jumping on the ground, "player" didin't jump in the rhytem
of the ball (the ball was not "regulating" behavior),
but he just waited and concentrated to catch it
with hands prepared (I suppose "player" was predicting or
estimating when the ball will fall into hands).

BP :
A control system with an integrating output function is needed to do this.
The first control system keeps the fielder in the (vertical) plane of the
ball's trajectory, while the second one then changes the running speed over
a range between positive and negative limits.
This sort of arrangement will in fact result in the fielder's
following a path that places him at the exact spot where the ball is
arriving.

HB :
So ifI understand right, you suppose that characteristics of flying ball
(*ffordance) are in every moment "regulating" the behavior of the player ?
And that some "control system with integrating output function in needed" to
do this ?
Isn't "NEW" theory (model) something that should describe the "physical
processes that the modelled system is carying out" (LCS III, p. 10), not
guessing what kind of integrating function should be "implanted" in control
system to match the theory ?

BP: It is rare because it works so poorly; it seldom results in
actually catching the ball. So if a control system is organized to
work that way, it will experience many large errors, and will be
reorganized until it works the right way.

HB :
What is the right way that players should work.
In accordance with "NEW" theory ?

BP :
Wanting to catch the ball is far from knowing where the ball will
come down. I strongly doubt that any successful fielder is organized that
way.

HB :
Now Bill I'm quite sure that you exactly know what is going on, and you are
as many times before "protecting" Rick's theory which have some holes in it.
I don't say it's bad model or something like that, I'm only trying to say
that it needs improvement.

BP: Sure, but that alone isn't enough to make him able to catch it.
The player is also controlling for "winning the game" even if his
team has lost 20 games in a row. A professional fielder probably
catches a high percentage of catchable balls, but the ony way to do
that is by using the best way to do it.

HB :
I agree that professional fielder is controlling many goals in the game and
his desire is to catch as many as possible balls. But I disagree that the
"ONLY WAY" to do succesful catching is the "NEW WAY". I think it's not only
up to player that he will be succesfull but also up to his position on the
field, whether he stands on the "better or worse" place in the relation with
the too
fast flying ball. But I'm sure that player will reorganize and do
the best to catch that ball if it's possible. So we get rare and incredible
"catchings" as the result of reorganization (inovation) of fielders who do
their job right. Would you be so kind and explain catching in attachment ?

BP: But you haven't offered an actual explanation, have you? You've
sketched in some of the things your model has to accomplish, but you
haven't said anything that would lead to an actual working model.

HB :
It's not my model. It's your model (or Ricks "old model", whatever) or two
theories you mentioned in the beginning. I'm just trying to say that "OLD
theory" is better one and in accordance with PCT as it has more chances to
explain any behavior in the physical world (baseball game or any other
sports game). Isn't it that what you keep saying in PCT about
"models and generalizations".

BP :
Of course we have the mathematical description. How do you think
those models that Rick and I work out end up actually working? The
programs contain the mathematical operations that we proposed to be
carried out by neural circuits. Rick has actual data from other
researchers, measurements obtained from cameras carried by fielders
as they run to catch balls, and other videos that show their paths
and speeds across the playing field. Those give us the numbers that
the model has to match by its own behavior. I'll let Rick show you
the programs he is using, since I'm just watching from the sidelines.

HB:
Oh wonderfull. I'd really like to see Rick's work. I'm speccially interested
in meassurments obtained from cameras on fielders and other videos and data
from other researchers.

BP: In the comparator, what happens is that the perceptual signal is rapidly
altered by the changes in running speed and direction, until the error gets
close to zero.

HB :
You described the perceptual signal. And what is reference ?

Best,

Boris

BEST BASEBALL CATCH EVER.flv (1.25 MB)

路路路

----- Original Message -----
From: "Bill Powers" <powers_w@FRONTIER.NET>
To: <CSGNET@LISTSERV.ILLINOIS.EDU>
Sent: Wednesday, November 21, 2012 1:58 AM
Subject: Re: Analyzing feedback paths

At 03:24 PM 11/20/2012 +0100, boris_upc wrote:

HB: O.K. I think that we need unique theory of ball-catching (human
behavior)

As far as I can see you are proposing two different theories:

"NEW" THEORY :
BP:

1. It is not necessary for the

catcher to estimate where the ball will come down. If the catcher moves
so
as to keep the apparent position of the ball vertically above home plate
at some constant or slowly changing angle, the catcher will find himself
exactly where the ball comes down, or close enough to catch it.

"OLD" THEORY :
BP :

2. I have satisfied myself by watching baseball on television that

catchers (more properly, "fielders") who try to catch balls in the way
the
old theory assumes usually fail to catch it. Sometimes they succeed, and
such successes are often remembered exactly because they are so rare.

BP: The "old" theory is the one most scientists have believed and which I
think is wrong. Some baseball players believe it, and when they try to
catch a ball that way (by running to the place they predict it will come
down) they usually can't catch it. The right way, and the one that just
about always works to catch a ball that is possible to catch, is the first
one. The player adjusts the direction and speed of running first so the
ball appears to be rising straight up (left-right angle remains constant
relative to objects on the ground), and then so it rises only very slowly
or stays at the same vertical visual angle relative to objects on the
ground.

Those are the two perceptions that have to be controlled by varying speed
and direction of running. It is easy to do and requires no predictions or
calculations about the path of the ball or where it will come down.
Players who consistently catch fly balls probably have organized
themselves the "New" way.

HB:
Both theories (if I understand right) explain special cases or maybe I can
say special "playing situations". And as you say they are experienced (on
TV), I can probably conclude that both are right, as they correctly
predict
or describe behavior of "real" players in "real" situation.
I suppose that only the theory which describes or predicts behavior in the
way
that matches the "reality" is for me right theory.

BP: But one of them doesn't work well, and does not normally result in
catching the ball. Very few successful fielders use it. This is because
it's impossible for a fielder to estimate accurately enough where a fly
ball is going to land. If the fielder does catch the ball is it only
because the random scatter of estimated landing points happens to include
the actual landing point and the fielder is lucky once in a while.

I recommend that you go outdoors with a ball and a friend and try this
out. If you have trouble catching balls and are trying to do it by
predicting the path of the ball, you will find that the constant-direction
works like magic, bringing you to the right place every time the ball can
actually be caught.

HB: In 1.case you say (if I understand right) that ball (*ffordance) is
"directing" behavior of the catcher (fielder), because the catcher
(fielder)
is "looking" at the ball all the time (with some constant or slowly
changing
angle) and fielder "miraculously" find himself "exactly where the ball
comes
down or close enough to catch it". I somehow missed PCT explanation of
this
"situation".

In case 1, the reference condition for one control system specifies that
the ball is to be stationary in the left-right direction. If the perceived
velocity is to the right, the negative error is converted into a reference
signal at a lower level for turning the direction of running toward the
left. A positive error results in changing direction to the right. When
the perceived direction equals the reference direction (zero left-right
velocity), the direction of the ball's trajectory is vertical, either
rising or falling. A control system with an integrating output function is
needed to do this.

For control in the up-down direction, a similar design is needed, with the
reference signal indicating an angle above the playing field, and the
error signal causing running toward the ball or away from the ball
depending on the sign of the error. The first control system keeps the
fielder in the (vertical) plane of the ball's trajectory, while the second
one then changes the running speed over a range between positive and
negative limits.

This sort of arrangement will in fact result in the fielder's following a
path that places him at the exact spot where the ball is arriving. No
magic or miracles are required.

I think you can see that if the ball is not moving either to the left or
the right, the fielder must be in the plane of the ball's arching
trajectory. When the ball lands, it will still be in that plane, but the
fielder might be too close or too far away.

The second control system keeps the apparent rate of change of the
vertical angle close to zero. If you draw a diagram showing the ball's
trajectory, with the player standing at any distance, you will see that
after the ball passes its maximum altitude and starts down again, the
direction of the vertical angle between the fielder's head and the ball
depends on whether the fielder is too close or too far away. If the
fielder is too far away, that angle will be decreasing, so the error
signal in that system should make the fielder run closer, and the opposite
if the fielder is too close. So both control systems are quite simple, and
when they both experience zero error when the ball comes down, the ball
will hit the fielder in his head, or in the dominant eye.

HB: In 2. case (if I understand right) you are saying that you saw on TV
how
catchers (fielders) are estimating the flight of the ball and run to the
place where
they can catch it. And these succesfull catching are very rare.

My oppinion is, that even if behavior is rare, it's part of the control
system which is functioning in the same way in any case (1. and 2. case)
and
any behavior control system produces that we can perceive or imagine is
valid and must be included in theory and explained with the same "tools",
if
the theory is right and good.

BP: It is rare because it works so poorly; it seldom results in actually
catching the ball. So if a control system is organized to work that way,
it will experience many large errors, and will be reorganized until it
works the right way.

HB: If PCT is believed to be right and unique theory of human behavior,
than I
suppose it has to explain any "real" behavior of control system in
succesfull way.
Although I suppose it's better one PCT way for explaining all behaviors
not two different ways (for ex. old and new).
I don't see how can "real" control system work in separate cases
differently
(in new and old way) if there is only one PCT and if "real situation" is
reference
for the theory to be right or wrong.

BP: You are forgetting about reorganization. Control systems are not wired
into the brain so they can work only one way. They are a result of
contininuous reorganization that eventually finds a way to reduce errors
to zero, or close enough. Only a few of the hierarchical control systems
are working when a person is born, and even those have to be reorganized
to work properly when interactions with the environment start.

HB: If you saw it on TV, than it doesn't matter how rare behavior is, it
is
something what happened "in "reality" and control system did it, so if the
theory is right it will succesfully explain what has happened.
So I think that succesfull explanation with the unique theory which
explain
ALL behaviors of control system, is right and good.

And PCT is supposed to be that kind of theory.

BP: It seems that you have been thinking that all these control systems
exist in the brain from the beginning of life. I assume that only the
potential for control systems is there at first -- the neural and muscular
building blocks from which control systems can be created. The actual
control systems we find in adults are, I assume, the result of long years
of reorganization with both successes and failures happening again and
again along the journey. Any organization is possible. Tom Bourbon once
showed that a control system can be built that imitates a
stimulus-response system. But stimulus-response systems can't seek goals
and they can't produce repeatable results in a changeable environment or
when unpredictable disturbances occur, as they always do. So that kind of
organization will be reorganized away and replaced by a control system
that can do those things.

I think that in both cases catcher (fielder) is estimating the
"catching-point" as the goal of every behavior of the fielder in any case
is to
"catch the ball".

Wanting to catch the ball is far from knowing where the ball will come
down. I strongly doubt that any successful fielder is organized that way.

In 1. case he is estimating the catching-point on the bases of "continuous
perception", as he has time to "watch the ball". The more he watch the
ball
the more precise he will estimate the "catching-point", and the more
chances
that trial of "catching the ball" will be succesfull.

OK, so you have chosen to believe the old theory, the one you will find in
most textbooks on motor control. If you can produce a model that works
that way and can handle things like winds that disturb the path of the
ball, tell us about it. You could be right, even if I doubt it. I'd like
to see what the necessary perceptions are, and the computations, and what
accuracy and speed of computing are required. I know what is required for
the other model, and it's pretty simple. Yours sounds immensely complex.

HB: In the 2. case as I see it, fielder hasn't enough time to precisly
estimate
the flight of the ball, because it's probably to fast or/and to high. So
he saw it for the moment and start running immediatelly as he estimated
that any part of the
second that he miss, the controlling of the "catching a ball" will be
impossible. So he starts running maybe "blindly" or by some "feeling" to
the
"catching-point".

But as I see it, the basis of "estimation of the flight of the ball" is
the
same in all behaviors, no matter how much time fielder has to estimate
it's
trajectory. The process of control that's going on in organism is the same
in any case, if PCT is right theory.

So in any case as I see it, fielder is controlling for "catching the ball"
in certain point. If you'll watch on TV again, I'm sure you will precisely
see that any player in any "playing situation" is controling for "catching
the ball".

BP: Sure, but that alone isn't enough to make him able to catch it. The
player is also controlling for "winning the game" even if his team has
lost 20 games in a row. A professional fielder probably catches a high
percentage of catchable balls, but the ony way to do that is by using the
best way to do it.

HB: So I think that at least we must have some unigue model (concept) that
will
explain all the behaviors of fielders if the theory is matching the
"reality" ? We don't need two theories.

I'll try with Fred's GAP-ACT concept to make a unigue concept (model),
which would
be able to explain every behavior of the catchers (fielders). I'll try in
this way :
If in your 1. case, "looking at the ball" (for ex. it's "angular
velocity") is the referenceof the fielder, from what will perception be
subtracted ?

BP: If the reference signal is set to zero, the angle from which the
perceived angle is subtracted is zero. This would be the case for the
system that controls angular velocity in the horizontal direction. For the
vertically-oriented angle, the reference signal would have a value
corresponding to whatever angle has been found to work well enough.

So as I see it, every behavior by Fred model (concept) can be explained,
so
the theory should be working for all goal-directed behaviors of human.

BP: But you haven't offered an actual explanation, have you? You've
sketched in some of the things your model has to accomplish, but you
haven't said anything that would lead to an actual working model.

HB: Can you use the same Freds' concept to explain your vision what is
happening. You can make also mathematical description of what is
happening.
I'm really interested to see it.

Of course we have the mathematical description. How do you think those
models that Rick and I work out end up actually working? The programs
contain the mathematical operations that we proposed to be carried out by
neural circuits. Rick has actual data from other researchers, measurements
obtained from cameras carried by fielders as they run to catch balls, and
other videos that show their paths and speeds across the playing field.
Those give us the numbers that the model has to match by its own behavior.
I'll let Rick show you the programs he is using, since I'm just watching
from the sidelines.

HB: I'm most interested in PCT explanation of this "new" one :

BP: Actually this is the old theory of ball-catching that Rick's
model and other related ones refute. It is not necessary for the
catcher to estimate where the ball will come down. If the catcher
moves so as to keep the apparent position of the ball vertically
above home plate at some constant or slowly changing vertical angle, the
catcher will find himself exactly where the ball comes down, or close
enough to catch it.

HB :
What excatly is happening in the comparator while catcher (fielder) "moves
so as to keep the apparent position of the ball.".
What is so apparent that it don't need PCT explanation ?\

BP: You'll find most of it earlier in this post. In the comparator, what
happens is that the perceptual signal is rapidly altered by the changes in
running speed and direction, until the error gets close to zero.

The word "apparent" was used as opposed to "actual". As the fielder moves,
the angle of the ball as the fielder sees it seems to change up, down,
left, and right, but that is only the way it looks. Its actual path
through space doesn't change that way. The apparent changes are caused by
the changing relationship between the fielder and the ball.

Best,

Bill P.

[From Fred Nickols (2012.11.20.0836 AZ)]

I'll try, Boris.

In terms of the GAP-ACT Model, the target variable (T) is the distance
between the fielder and the ball. The goal (G) is for that distance to be
sufficiently small so as to enable catching the ball. The actions (A)
consist of running, altering direction and speed, keeping an eye on the
ball, monitoring the distance reduction, etc. The perceptions of interest
(P) are the distance between the fielder and the ball and the rate of
closing any gap. If the distance is increasing despite the fielder's best
efforts, the ball is not going to be caught, at which point the fielder
might do something like predict where the ball will hit and how it might
bounce (e.g., off the outfield wall or into it) and he will head for that
point. The conditions (C) might include the wind (causing the ball to
drift > but compensated for by the running), and the surface of the
outfield
(e.g., the ball takes a weird bounce and I think most of us have seen
outfielders > surprised by that "disturbance").

FN (earlier) :
The relative movement between me (one the run) and the ball
(on the fly) tell me if I'm likely to intercept it or not. If so, I keep
doing what I'm doing. If not, I speed up, change course, slow down or
whatever it takes to maintain my sense that I am likely to intercept the
ball. I do not estimate where it will land; I estimate or take stock of the
likelihood that I can/will intercept it.

FN (earlier) :
...I think the fielder/catcher is predicting whether
or not he can intercept the ball....

HB :
I see some problems in your description of experiences with baseball game
and GAP-ACT "explanation" :

聽聽1.. I think that T (target) can't be the distance between fielder and the
ball, because it looks like something "objective"
in the environment. I see it as description from the position of the
observer. I think it has to be in perceptual terminology, how you perceive
the distance to the ball, and what is happening in comparator. Your
assumption is probably that fielder (you) controls for the distance between
the fielder and the ball (T). I don't know how it's possible to control for
distance between you and ball, as you can't see yourself. Maybe on video :).
聽聽2.. It also seems problematic to me, that your assumption is that fielder
must see the ball all the time if he wants to "control for the reducing of
the distance" (G). So you can analyze only behaviors with that
characteristics. Other behaviors are "out of sight" . So I think that your
description of GAP-ACT "model" can't explain all the behaviors in baseball
game, as it don't include behaviors in which fielder don't look at the ball.
Where the fielder is running, when he is not looking at the ball ? He can't
see the ball, so the G can't be distance any more. What it is then ?
聽聽3.. In your description of experiences with the baseball game it seems to me
that you emphasize that the distance between you and the ball tells you
whether you will intercept (catch) the ball or not. So the reference is
intercepting (catching the ball) as you "regulated" your behavior in
relation to whether you were "likely to intercept it or not". It seems that
"distance between you and the ball" could be some lower level controlled
perception that "helped" you to reach the goal : intercepting the ball.

I think that any "theory" that supports "all the time visual contact of the
fielder with the ball" can'f explain all the behaviors that appear in the
baseball game, so it's not good theory.

FN :

Again, please keep in mind, Boris, that what I am offering is not a model
of > outfielder behavior; I'm simply examining what I can observe and
explaining > it in terms of the GAP-ACT (PCT) framework. My explanations
could be > right> on the money or woefully misbegotten.

Fred Nickols

HB :
Your obervations and examining were welcome. It's also possible
that I misunderstood something, you know my English..

So I have one request. Can you translate this for me please : "right on the
money or woefully misbegotten".

Best,

Boris

路路路

----- Original Message ----- From: "Fred Nickols" <fred@NICKOLS.US>
To: <CSGNET@LISTSERV.ILLINOIS.EDU>
Sent: Tuesday, November 20, 2012 4:46 PM
Subject: Re: Analyzing feedback paths

I'll try, Boris.

In terms of the GAP-ACT Model, the target variable (T) is the distance
between the fielder and the ball. The goal (G) is for that distance to be
sufficiently small so as to enable catching the ball. The actions (A)
consist of running, altering direction and speed, keeping an eye on the
ball, monitoring the distance reduction, etc. The perceptions of interest
(P) are the distance between the fielder and the ball and the rate of
closing any gap. If the distance is increasing despite the fielder's best
efforts, the ball is not going to be caught, at which point the fielder
might do something like predict where the ball will hit and how it might
bounce (e.g., off the outfield wall or into it) and he will head for that
point. The conditions (C) might include the wind (causing the ball to
drift
but compensated for by the running), and the surface of the outfield
(e.g.,
the ball takes a weird bounce and I think most of us have seen outfielders
surprised by that "disturbance").

Again, please keep in mind, Boris, that what I am offering is not a model
of
outfielder behavior; I'm simply examining what I can observe and
explaining
it in terms of the GAP-ACT (PCT) framework. My explanations could be
right
on the money or woefully misbegotten.

Fred Nickols

-----Original Message-----
From: Control Systems Group Network (CSGnet)
[mailto:CSGNET@LISTSERV.ILLINOIS.EDU] On Behalf Of boris_upc
Sent: Tuesday, November 20, 2012 8:31 AM
To: CSGNET@LISTSERV.ILLINOIS.EDU
Subject: Re: Analyzing feedback paths

Hi Fred,

what a nice surprise...Tahnks for your answers :slight_smile:

[Fred Nickols] I don't. I think the fielder/catcher is predicting

whether or not

he can intercept the ball. He's not focused on the physical landing
point

but,

rather, the relative change in position between himself and the ball.

Fred Nickols] Close but not quite. I don't think the catcher is

estimating

> the catching point. He's keeping track of how well he's reducing the
> distance between himself and the ball, not where the ball will likely
> come down.

[Fred Nickols] I don't. I think the fielder/catcher is predicting
whether
> or not he can intercept the ball. He's not focused on the physical
> landing point but, rather, the relative change in position between
> himself and the ball.

The initial trajectory
> tells me something too; namely, the general direction, movement and
speed.
> Depending on what that is, I take off running, all the while keeping
> an eye on the ball. The relative movement between me (one the run)
> and the ball (on the fly) tell me if I'm likely to intercept it or
> not. If so, I keep doing what I'm doing. If not, I speed up, change
> course, slow down or whatever it takes to maintain my sense that I am
> likely to intercept the ball. I do not estimate where it will land; I
> estimate or take stock of the likelihood that I can/will intercept it.

HB:
Nice descriptions of the baseball game. Could you now explain everything
what you wrote with GAP_ACT model orconcept of whatever... or PCT model

Best,

Boris

----- Original Message -----
From: "Fred Nickols" <fred@NICKOLS.US>
To: <CSGNET@LISTSERV.ILLINOIS.EDU>
Sent: Tuesday, November 20, 2012 3:49 PM
Subject: Re: Analyzing feedback paths

> [From Fred Nickols (2012.11.20.0733 AZ)]
>
> My comments are embedded below. Material to which I am not
responding has
> been snipped.
>
>> -----Original Message-----
>> From: Control Systems Group Network (CSGnet)
>> [mailto:CSGNET@LISTSERV.ILLINOIS.EDU] On Behalf Of boris_upc
>> Sent: Tuesday, November 20, 2012 7:25 AM
>> To: CSGNET@LISTSERV.ILLINOIS.EDU
>> Subject: Re: Analyzing feedback paths
>>
>> O.K. I think that we need unique theory of ball-catching (human

behavior)

>
> [Fred Nickols] I think we have a theory of ball-catching, as expressed

in

> Rick Marken's model.
>>
>> "OLD" THEORY :
>> BP :
>>
>> 2. I have satisfied myself by watching baseball on television that
>> > catchers (more properly, "fielders") who try to catch balls in the

way

>> > the old theory assumes usually fail to catch it. Sometimes they
>> > succeed, and such successes are often remembered exactly because
they
>> are so rare.
>
> [Fred Nickols] Actually, I don't think Bill can say what he wrote

above.

> As he likes to point out, we can't tell what someone is up to by simply
> observing their behavior. He might "guess" that some fielders are

trying

> to
> estimate where the ball will come down but I don't think he can really
> tell
> that's what they're trying to do.
>
>>
>> HB:
>> Both theories (if I understand right) explain special cases or maybe I
>> can
> say
>> special "playing situations". And as you say they are experienced (on
>> TV),
> I
>> can probably conclude that both are right, as they correctly predict
>> or
>> describe behavior of "real" players in "real" situation.
>> I suppose that only the theory which describes or predicts behavior in
>> the
>> way that matches the "reality" is for me right theory.
>>
>> In 1.case you say (if I understand right) that ball (*ffordance) is
> "directing"
>> behavior of the catcher (fielder), because the catcher (fielder) is
> "looking" at
>> the ball all the time (with some constant or slowly changing
>> angle) and fielder "miraculously" find himself "exactly where the ball
> comes
>> down or close enough to catch it". I somehow missed PCT explanation of
> this
>> "situation".
>
> [Fred Nickols] Again, I think that's Rick's model.
>>
>> I think that in both cases catcher (fielder) is estimating the
> "catching-point" as
>> the goal of every behavior of the fielder in any case is to "catch
>> the
> ball".
>
> [Fred Nickols] I'm not a good ball player but I have played some and in
> the
> outfield. The first thing to which I recall responding is the crack of
> the
> bat hitting the ball. The sound tells me something. The initial
> trajectory
> tells me something too; namely, the general direction, movement and
speed.
> Depending on what that is, I take off running, all the while keeping an
> eye
> on the ball. The relative movement between me (one the run) and the

ball

> (on the fly) tell me if I'm likely to intercept it or not. If so, I

keep

> doing what I'm doing. If not, I speed up, change course, slow down or
> whatever it takes to maintain my sense that I am likely to intercept
> the
> ball. I do not estimate where it will land; I estimate or take stock
> of
> the
> likelihood that I can/will intercept it.
>
>> We don't need two theories.
> [Fred Nickols]
> I don't think Bill was advancing two theories but I'll leave that to

him.

>>
>> I'll try with Fred's GAP-ACT concept to make a unigue concept (model),
> which
>> would be able to explain every behavior of the catchers (fielders).

I'll

> try in
>> this way :
>>
>> INSIDE ORGANISM
>> G = goal (wanted perception, catching the ball or the ball in the
>> glove
>> of
> the
>> catcher) A = action (internal environment. I suppose that Fred ment,
>> if

I

> see
>> right, nervous-muscles connection, driven by "error" in comparator) P
>> =
>> perception (I suppose perception of the ball)
>>
>> OUTSIDE ORGANISM
>> A = action (behavior, outer expression of the muscles activity in
>> outer
>> environment that can be observed) C = conditions (probably Fred meant
>> variables in environment that affect the ball - disturbances - and

other

>> environmental variables) T = target (I suppose he meant controlled
> variable
>> in outer environment, the
>> ball)
>
> [Fred Nickols] Fred meant "ball in glove" or, in shorthand, "caught"
>>
>> So my opppinion is that the catcher (fielder) is perceiving the flight

of

> the
>> ball, and in imagination starts to estimate the place where the ball

will

> fall.
>
> [Fred Nickols] I don't. I think the fielder/catcher is predicting

whether

> or not he can intercept the ball. He's not focused on the physical
> landing
> point but, rather, the relative change in position between himself and

the

> ball.
>
>> There are probably differences in how fast "players" estimates the
> "catching
>> point".
>>
>> When the goal is set (wanted perception of the ball), the difference
>> between goal and actual flight of the ball is perceived, so with
>> action
> the
>> fielder starts to reduce the difference between actual perception of

the

>> flying ball and reference - estimated catching point in imagination.
>
> [Fred Nickols] Close but not quite. I don't think the catcher is
> estimating
> the catching point. He's keeping track of how well he's reducing the
> distance between himself and the ball, not where the ball will likely

come

> down.
>
>>
>> If in your 1. case, "looking at the ball" (for ex. it's "angular
> velocity") is the
>> reference of the fielder, from what will perception be subtracted ?
>>
>> So as I see it, every behavior by Fred model (concept) can be

explained,

> so
>> the theory should be working for all goal-directed behaviors of human.
>>
>> Can you use the same Freds' concept to explain your vision what is
>> happening. You can make also mathematical description of what is
>> happening.
>> I'm really interested to see it.
>>
> [Fred Nickols] Boris: I'm glad you like my GAP-ACT model but I don't
> think
> it's quite fair to ask Bill to explain anything in terms of it.

Besides,

> it's not "my" concept; it's simply a view based on PCT.
>

[From Fred Nickols (2012.11.26.0645 AZ)]

HB :
Your obervations and examining were welcome. It's also possible that I
misunderstood something, you know my English..

So I have one request. Can you translate this for me please : "right on

the

money or woefully misbegotten".

Best,

Boris

[Fred Nickols] "Right on the money" is a phrase meaning "exactly right" or
"correct." The British say "Spot on."
"Woefully misbegotten" means "very poorly conceived" referring to an idea or
suggestion or plan or course of action with absolutely no merit.
At least that's what I mean by those two terms.

Fred Nickols

[From Bill Powers (2012.11.26.1010 MST)]

BP earlier: The "old" theory is the one most scientists have believed and
which I think is wrong.

HB :
You think it's wrong ? It seems to me, that you are pretty unsure about that
? Who are those "most" scientist ?

BP: Yes, I do need someone else to tell me I'm not crazy. This wouldn't be the first idea I got wrong. The "most scientists" are the ones whose words I have read who assumed that ball catchers estimate where the ball will come down and run to that place. I have rarely seen any other theory offered. The first time I saw a different theory was in a 1960's article in Science magazine by a physicist named Chapman, who proposed that the ball catcher moves so as to keep the tangent of the angle from horizon to ball slowly increasing. I tried it out and agreed with him, though the "tangent" idea didn't seem necessary. Just keeping the angle slowly increasing worked, and I soon found that keeping it constant also worked. Since then there have been others who offered somewhat different definitions of the controlled variable; Rick Marken has been working with some of them to apply the test for the controlled variable to pick out the right one. The one that works best is basically Chapman's idea, somewhat modified.

HB: Is the "old theory" - "running to the place they predict the ball will fall
down" ? Do I understand right that the "old theory" has for goal "catching
the ball" or "intercepting it" ?

BP: Yes, that's the overall goal in any case, to catch the ball. That's not the issue. The issue is how the fielder manages, when the ball comes down, to get to the right place so as to be near enough to catch it. Certainly, if the fielder could accurately and quickly estimate where the ball would come down, he could run to that place and catch the ball. But doing it that way requires very precise perceptions, a lot of difficult mental mathematics, and very precise execution of the running part of the task. Since there is a far simpler way to get to the right place in time to catch the ball, one that requires no estimates of where the ball will come down, I have found it easier to believe that the simple way is the one that a skillful fielder will learn to use.

HB: I thought this is a PCT theory where reference is something "presumably
determined in your brain" (LCS III, p. 21). So predicting the place where
the ball will fall down or "the catching point" or "interecepting point" is
something "presumably determined in our brain".

If it were done that way, yes, it would have to be done in the brain. I doubt that anyone can actually do it that way. Here is a link to a discussion:

http://www.livescience.com/3445-baseball-players-catch-fly-balls.html

Rick Marken has proven that Chapman's "Optical Acceleration Cancellation" theory (supported by a modern researcher) is not the best one; keeping a constant velocity works much better in a model. The model predicts the running path of the fielder within a few centimeters. But it's interesting that modern researchers are indeed thinking of controlled variables and no longer believe that calculating the future path of the ball is a likely explanation.

I like your "NEW" definition of reference signal :
"The intended value of the perceptual signal is indicated by another
signal of exactly the same kind as the perceptual signal, but with a
different source. This other signal, the reference signal, is an example of
the perceptual signal's state as it would be when the controlled variable is
in its intended state. (LCS III, p.31).

This seems to me perfect description of goal, which could be "catching" or
"intercetping" the ball as some "intended" state (catching
point) is "construction" in the brain...

That's not a precise enough definition of the goal for use in a model. The model has to demonstrate that it follows a path like the one the real fielder follows.

BP :
Some baseball players believe it, and when they try to catch a ball that way
(by running to the place they predict it will come down) they usually can't
catch it.

HB :
Well this one will be hard to prove. Do you really think that some
rare behaviors are "attached" to specific baseball players ?
Do you maybe know who are those some baseball players
that "believe in rare behaviors" ?

Yes, that's what I think. Coaches may tell players to guess where the ball will come down and run to that place, and players who believe the coach may try to do it that way. But they will not be very good at catching balls. The players who are best will be those who use the simple and accurate way, which is to run so the ball appears to be almost stationary in the sky.

Maybe one name ? Can Rick give some names as he has so much
material upon baseball catch as you wrote in the end of your post ?

BP: Yes, he is in contact with many of the modern researchers in this field of study. He can supply quite a few names. Rick?

BP : The right way, and the one that just about always works to catch a ball
that is possible to catch, is the first one..//.Those are the two
perceptions that have to be controlled by varying speed and direction of
running.
It is easy to do and requires no predictions or calculations about the path
of the ball or where it will come down. Players who consistently catch fly balls
probably have organized themselves the "New" way.

HB :
Does "two perceptions that have to be controlled" means that players are
observing the ball all the time and ball is determining players behavior in
every moment ?

Yes, to the question about the players observing the ball all (or nearly all) of the time. No, this is not a case of the ball determining the player's movement. That is what we call the "behavioral illusion." The player is varying his running direction and speed so as to keep the ball's apparent position in the sky nearly constant (after the ball begins to come down). As a consequence of doing this, the player will arrive at the place where the ball comes down just as the ball gets there -- and without any predictions or complex computations.

What does it mean "PROBABLY" have organized themselves the "NEW way" ?

I mean it would be reasonable to assume that the fielders are moving so as to keep the ball's apparent position in the sky almost stationary. One would have to ask the player how he does it to be sure in any given case. Some players may have learned a different way.

BP: But one of them doesn't work well, and does not normally result
in catching the ball. Very few successful fielders use it. This is
because it's impossible for a fielder to estimate accurately enough
where a fly ball is going to land. If the fielder does catch the ball
is it only because the random scatter of estimated landing points
happens to include the actual landing point and the fielder is lucky
once in a while.

HB:
Does this mean that behaviors which are resulting in "rare and few sucessful
trials" should be excluded from "NEW" theory ? Can "NEW" theory explain those
"rare" behaviors, speccially those when fielder is not observing the ball ?

No, but a model can, if the experiments with real fielders don't convince you that the old explanation is wrong. A model could be programmed to find out just how accurate the perceptions, computations, and motor actions would have to be to explain the observed behavior, and I would predict that this would show why the old method works so poorly.

BP :
I recommend that you go outdoors with a ball and a friend and try
this out. If you have trouble catching balls and are trying to do it
by predicting the path of the ball, you will find that the
constant-direction works like magic, bringing you to the right place
every time the ball can actually be caught.

HB: Although throws were easy, we have much troubles with
getting used to hitting and catching the ball. So we'll try to repeat it.
But one thing is sure. If ball was hit toward someone,
and it was jumping on the ground, "player" didin't jump in the rhytem
of the ball (the ball was not "regulating" behavior),

BP: Correct, the ball does not regulate the behavior. Did you expect it to do that?

HB: but he just waited and concentrated to catch it
with hands prepared (I suppose "player" was predicting or
estimating when the ball will fall into hands).

BP: Did you throw the ball toward the player, or to a different place so the player had to run to catch it?

HB :
So if I understand right, you suppose that characteristics of flying ball
(*ffordance) are in every moment "regulating" the behavior of the player ?
And that some "control system with integrating output function in needed" to
do this ?

I don't know where you get the idea that the ball regulates the behavior. The fielder varies his own behavior to keep the ball looking as if it is stationary in the sky (as projected on the sky -- it is, of course, approaching the fielder).

Isn't "NEW" theory (model) something that should describe the "physical
processes that the modelled system is carying out" (LCS III, p. 10), not
guessing what kind of integrating function should be "implanted" in control
system to match the theory ?

Yes, the new theory does exactly that. But the integrating function is necessary to achieve stability and accuracy.

BP: It is rare because it works so poorly; it seldom results in
actually catching the ball. So if a control system is organized to
work that way, it will experience many large errors, and will be
reorganized until it works the right way.

HB :
What is the right way that players should work.
In accordance with "NEW" theory ?

Yes. That is the simplest way, the easiest way to learn, and the way that works best. Of course you can try any method you please, but I predict that the "NEW" way will still work best.

BP :
Wanting to catch the ball is far from knowing where the ball will
come down. I strongly doubt that any successful fielder is organized that
way.

HB :
Now Bill I'm quite sure that you exactly know what is going on, and you are
as many times before "protecting" Rick's theory which have some holes in it.
I don't say it's bad model or something like that, I'm only trying to say
that it needs improvement.

No, it doesn't. It's your understanding of the theory that is insufficient.

BP earlier: Sure, but that alone isn't enough to make him able to catch it.
The player is also controlling for "winning the game" even if his
team has lost 20 games in a row. A professional fielder probably
catches a high percentage of catchable balls, but the ony way to do
that is by using the best way to do it.

HB :
I agree that professional fielder is controlling many goals in the game and
his desire is to catch as many as possible balls. But I disagree that the
"ONLY WAY" to do succesful catching is the "NEW WAY".

BP: OK, that's clear enough. How about doing some tests of different ways of catching the ball, and see which way works best? All the ways that I know of that others have proposed involve more complexity, more need for rapid mental calculations, and more need for precise predictions compared to what I call the new way. And experiments with fielders who try to predict where a ball will come down show that even professional baseball players are not very good at doing that. But you could be right, so demonstrate that you are.

HB: I think it's not only
up to player that he will be succesfull but also up to his position on the
field, whether he stands on the "better or worse" place in the relation with the too
fast flying ball. But I'm sure that player will reorganize and do
the best to catch that ball if it's possible. So we get rare and incredible
"catchings" as the result of reorganization (inovation) of fielders who do
their job right. Would you be so kind and explain catching in attachment ?

BP: I did explain it. If the player runs to where he thinks the ball will come down, his estimate will have some errors in it. Occasionally, the prediction will happen to be close to the actual landing point, so some of the fly balls will be caught. But that is a matter of luck, not skill. According to my observations of fielders in action, those who turn and run to the place they think the ball will come down (not many, but I have seen some do it that way) usually fail to catch the ball. Very rarely, they will catch it, but only by chance.

BP: But you haven't offered an actual explanation, have you? You've
sketched in some of the things your model has to accomplish, but you
haven't said anything that would lead to an actual working model.

HB :
It's not my model. It's your model (or Ricks "old model", whatever) or two
theories you mentioned in the beginning. I'm just trying to say that "OLD
theory" is better one and in accordance with PCT as it has more chances to
explain any behavior in the physical world (baseball game or any other
sports game). Isn't it that what you keep saying in PCT about
"models and generalizations".

BP: I see that you have been trying, all through this post, to defend the old theory! I didn't understand that before this point. Well, if you want to go on believing in it, there is nothing I can do about that. I can offer sympathy because I, too, have believed ideas that turned out to be untrue, and I didn't like the experience. I wish you luck in finding a way to prove that the old theory is the better one. However, I advise you to be prepared for disappointment.

HB: Oh wonderfull. I'd really like to see Rick's work. I'm specially interested
in meassurments obtained from cameras on fielders and other videos and data
from other researchers.

Rick, can you supply Boris with some of these materials?

BP: In the comparator, what happens is that the perceptual signal is rapidly
altered by the changes in running speed and direction, until the error gets
close to zero.

HB :
You described the perceptual signal. And what is reference ?

The controlled variable is a perception of the changing angular relationship of the ball to the objects in a frame of reference anchored to the playing field. The reference condition is a ball in an almost stationary angular relationship to those objects, as seen from the fielder's changing position. In the models, perceptual and reference variables are expressed in terms of angles in the left-right and the vertical directions (one control system for each direction).

Best,

Bill P.

路路路

At 10:24 AM 11/26/2012 +0100, boris_upc wrote:

[From Rick Marken (2012.11.26.2145)]

Bill Powers (2012.11.26.1010 MST) to Boris Hartman

BP: Rick Marken has proven that Chapman's "Optical Acceleration Cancellation"
theory (supported by a modern researcher) is not the best one; keeping a
constant velocity works much better in a model. The model predicts the
running path of the fielder within a few centimeters.

RM: And it does it when the objects being caught (intercepted) have
highly unpredictable trajectories, like Frisbees and toy helicopters.

BH: Maybe one name ? Can Rick give some names as he has so much
material upon baseball catch as you wrote in the end of your post ?

BP: Yes, he is in contact with many of the modern researchers in this field
of study. He can supply quite a few names. Rick?

Here are the names of some of the leaders in the field:

Michael McBeath
Dennis Shaffer
Peter McLoed
J. R. Tresilian
W, H. Warren
R. S. Marken :wink:

HB: Oh wonderfull. I'd really like to see Rick's work. I'm specially
interested in meassurments obtained from cameras on fielders
and other videos and data from other researchers.

BP: Rick, can you supply Boris with some of these materials?

RM: I have many spreadsheets full of "object interception" data but
all this data was collected by others and sent to me for analysis and
I don't think I'd feel right about distributing it without their
permission. And I wouldn't even ask for permission unless I was
convinced that I wouldn't be wasting my time. It would take me some
time to explain the data and the models implemented in the spreadsheet
and I really don't want to waste my time explaining it to people who
who in my estimation are neither willing nor able to understand it.

Best regards

Rick

路路路

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

Hi Bill and Rick,

I must admitt that I expected something like this :slight_smile:

Well never mind. I still think that your model needs quite some
improvements.

I'll just try to comment "best catch ever" which I sent in attachement.
I think that fielder couldn't catch that ball according to "NEW" way or
shall we say "in best way".
So I think he came to the limits of "NEW" theory, and started to use
"OLD" one.
So the player in video has to improvize (reorganize) if he wanted to score
as there was no "normal way" to catch the ball. He has to invent something.

After all he is professionall and he has to do any effort for his team.
So when it was obvious that catching in "your way, the best way" was
impossible as the ball was flying to high into the encloser, he has to find
some real new way to catch it. And there it was. The best solution ever.
As there was no ledder, he has to use his own skills. Isn't it wonderfull
to see the limits of human abilities ?

And as I see it usually mathematical theories with highly predictable and
limited "patterns" of behavior, fail when human inovation, creativty
(reorganization) is needed. I think mathematical model will never be able to
excatly predict human behaviors in the game.

But as I said before and I really beleive it, that you are pioneers of this
task, although I doubt that we'll be able to see the whole results of your
effort in life-time. As I see it, technology hasn't come to the point that
it could consider all the knowledge that PCT can offer. There so much to be
done. I think that those who tried, understand what I'm saying.

And this is my oppinion about baseball catch. When the players come to the
limits of "NEW" theory (I think they try to use it always, as I think that
they were trained in that way.....if the coach is good), and they see that
"NEW" theory doesn't work any more or estimate that it will not work in
actual situation, because of the special characteristics of playing
situation, they start to "use OLD theory".

I think that players train much of possible situations and catches in the
"right" way as you described it.But unpredictable situations in play
(their actual standing point,
characteristics of hits and flying ball, ability to predict possible trajectory
of the ball in the moment of hit, positions of other
players.etc.) demand from players also improvization (reorganizations) of
special kind.

I think that Fred, when decribing his experiences in baseball game,
wrote something what I'm trying to say. Only that he has much more.
He has his personal experience.

FN :
The relative movement between me (one the run) and the ball
(on the fly) tell me if I'm likely to intercept it or not. If so, I keep
doing what I'm doing. If not, I speed up, change course, slow down or
whatever it takes to maintain my sense that I am likely to intercept the
ball....I think the fielder/catcher is predicting whether
or not he can intercept the ball....

HB :
I think it's like Bill said, it's up to us. It's the purpose of the user...

Best,

Boris

路路路

----- Original Message ----- From: "Richard Marken" <rsmarken@GMAIL.COM>
To: <CSGNET@LISTSERV.ILLINOIS.EDU>
Sent: Tuesday, November 27, 2012 6:44 AM
Subject: Re: Analyzing feedback paths

[From Rick Marken (2012.11.26.2145)]

Bill Powers (2012.11.26.1010 MST) to Boris Hartman

BP: Rick Marken has proven that Chapman's "Optical Acceleration
Cancellation"
theory (supported by a modern researcher) is not the best one; keeping a
constant velocity works much better in a model. The model predicts the
running path of the fielder within a few centimeters.

RM: And it does it when the objects being caught (intercepted) have
highly unpredictable trajectories, like Frisbees and toy helicopters.

BH: Maybe one name ? Can Rick give some names as he has so much
material upon baseball catch as you wrote in the end of your post ?

BP: Yes, he is in contact with many of the modern researchers in this
field
of study. He can supply quite a few names. Rick?

Here are the names of some of the leaders in the field:

Michael McBeath
Dennis Shaffer
Peter McLoed
J. R. Tresilian
W, H. Warren
R. S. Marken :wink:

HB: Oh wonderfull. I'd really like to see Rick's work. I'm specially
interested in meassurments obtained from cameras on fielders
and other videos and data from other researchers.

BP: Rick, can you supply Boris with some of these materials?

RM: I have many spreadsheets full of "object interception" data but
all this data was collected by others and sent to me for analysis and
I don't think I'd feel right about distributing it without their
permission. And I wouldn't even ask for permission unless I was
convinced that I wouldn't be wasting my time. It would take me some
time to explain the data and the models implemented in the spreadsheet
and I really don't want to waste my time explaining it to people who
who in my estimation are neither willing nor able to understand it.

Best regards

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

Hi Fred,

I'm greatfull to you and I'm honered, that you shared your baseball
experiences with me.

I'm also thankfull to you for your translation and explanation...

Best,

Boris

路路路

----- Original Message ----- From: "Fred Nickols" <fred@NICKOLS.US>
To: <CSGNET@LISTSERV.ILLINOIS.EDU>
Sent: Monday, November 26, 2012 2:53 PM
Subject: Re: Analyzing feedback paths

[From Fred Nickols (2012.11.26.0645 AZ)]

HB :
Your obervations and examining were welcome. It's also possible that I
misunderstood something, you know my English..

So I have one request. Can you translate this for me please : "right on

the

money or woefully misbegotten".

Best,

Boris

[Fred Nickols] "Right on the money" is a phrase meaning "exactly right"
or
"correct." The British say "Spot on."
"Woefully misbegotten" means "very poorly conceived" referring to an idea
or
suggestion or plan or course of action with absolutely no merit.
At least that's what I mean by those two terms.

Fred Nickols

[From Fred Nickols (2012.11.27.0942 AZ)]

I feel the need to clarify my comments which you cite below, Boris.

From: Control Systems Group Network (CSGnet)
[mailto:CSGNET@LISTSERV.ILLINOIS.EDU] On Behalf Of boris_upc
HB:
I think that Fred, when decribing his experiences in baseball game, wrote
something what I'm trying to say. Only that he has much more.
He has his personal experience.

FN :
The relative movement between me (one the run) and the ball (on the fly)
tell me if I'm likely to intercept it or not. If so, I keep doing what

I'm doing. If

not, I speed up, change course, slow down or whatever it takes to maintain
my sense that I am likely to intercept the ball....I think the

fielder/catcher is

predicting whether or not he can intercept the ball....

[Fred Nickols] I think it's important to note that I am NOT predicting
where the ball will land or even where it is going. I am predicting the
likelihood that I can intercept it (based on my perception of the relative
movement between me and the ball). If it becomes clear to me that I can't
catch it I will likely change course and head toward where I DO predict the
ball is headed or will go (and I have often been wrong and surprised on that
score, especially when the ball takes an odd bounce).

Note also, that I am a rank amateur when it comes to playing ball. I have
no idea what the pros do.

Regards,

Fred Nickols, CPT
Managing Partner
Distance Consulting LLC
The Knowledge Worker's Tool Room
Blog: Knowledge Worker Tools
www.nickols.us | fred@nickols.us

路路路

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

From Bill Powers (2012.11.27.1350 MST)]

Fred Nickols (2012.11.27.0942 AZ) --

聽聽[Fred Nickols] I think it's important to note that I am NOT predicting

where the ball will land or even where it is going. I am predicting the
likelihood that I can intercept it (based on my perception of the relative
movement between me and the ball). If it becomes clear to me that I can't
catch it I will likely change course and head toward where I DO predict the
ball is headed or will go (and I have often been wrong and surprised on that
score, especially when the ball takes an odd bounce).

When you look at the ball and see that it is moving neither up, down, left, or right, but remains at a fixed angle in the sky, and when you see its image is getting larger, I think you will judge that you are going to intercept it. If it moves upward, you have to move away from it faster to bring it downward again; if it moves downward, toward it by moving a little slower away, or faster toward it. If it moves left, you move right. If right, you move left.

That is how the constant-angle model works, too. It requires no prediction and no guessing as to where the ball will land. You are not intentionally running to a particular place -- just keeping that ball at a constant angle. What could be simpler?

Boris knows what he believes and he intends to go on helieving it regardless of logic or data. I will stop trying to change his mind. What do you think of this model?

Best,

Bill P.

[From Fred Nickols (2012.11.27.1409 AZ)]

Keeping the ball at a constant angle and getting bigger is a fit with my
experience. An inability to do that suggests I ain't gonna catch it and
then I'm left guessing where the ball will go.

Fred Nickols

From: Control Systems Group Network (CSGnet)
[mailto:CSGNET@LISTSERV.ILLINOIS.EDU] On Behalf Of Bill Powers
Sent: Tuesday, November 27, 2012 1:58 PM
To: CSGNET@LISTSERV.ILLINOIS.EDU
Subject: Re: Analyzing feedback paths

From Bill Powers (2012.11.27.1350 MST)]

Fred Nickols (2012.11.27.0942 AZ) --

聽聽[Fred Nickols] I think it's important to note that I am NOT predicting
>where the ball will land or even where it is going. I am predicting
>the likelihood that I can intercept it (based on my perception of the
>relative movement between me and the ball). If it becomes clear to me
>that I can't catch it I will likely change course and head toward where
>I DO predict the ball is headed or will go (and I have often been wrong
>and surprised on that score, especially when the ball takes an odd

bounce).

When you look at the ball and see that it is moving neither up, down,

left, or

right, but remains at a fixed angle in the sky, and when you see its image

is

getting larger, I think you will judge that you are going to intercept it.

If it

moves upward, you have to move away from it faster to bring it downward
again; if it moves downward, toward it by moving a little slower away, or
faster toward it. If it moves left, you move right. If right, you move

left.

That is how the constant-angle model works, too. It requires no prediction
and no guessing as to where the ball will land. You are not intentionally
running to a particular place -- just keeping that ball at a constant

angle. What

could be simpler?

Boris knows what he believes and he intends to go on helieving it

regardless

路路路

-----Original Message-----
of logic or data. I will stop trying to change his mind.
What do you think of this model?

Best,

Bill P.

From Bill Powers (2012.11.27.1350 MST)]

Rick Marken(2012.11.27) --

Rick, do you have ball-catching data that gives the x,y,z positions of the ball and the fielder's head (or the camera) during the pursuit and catch? If so, you can compute the compass direction angle from fielder to ball, and the elevation angle above the horizontal plane for each iteration.

Let x be position of an object relative to 0,0,0 in the left-right direction, y be the position in the up-down direction, and z be the position in the direction at right angles to the x and y axes.

b = ball, f = fielder:

Lateral angle = arctan{(x[b] - x[f])/(z[b] - z[f])}

Vertical angle = arctan{(y[b] - y[f])/(z[b] - z[f])}

distance, b to f: SQRT{sqr(x[b] - x[f]) + sqr(y[b] - y[f]) + sqr(z[b] - z[f])}

Skip computing the arctan in the case where z[b] = z[f] where the angle would be 90 degrees.

You may already be computing at least the lateral and vertical angles in the model program.

You could also compute the first and second differences to look at the velocity and acceleration of the angles. From the first differences of the fielder's position you can get his direction and speed of movement.

You could show us the results of these computations without revealing the data gathered by other people.

Best,

Bill P.

[From Rick Marken (2012.11.27.1625)]

Bill Powers (2012.11.27.1350 MST)--

BP: Rick, do you have ball-catching data that gives the x,y,z positions of the
ball and the fielder's head (or the camera) during the pursuit and catch?

RM: You betcha!

BP: If so, you can compute the compass direction angle from fielder to ball, and
the elevation angle above the horizontal plane for each iteration.

RM: That's what I've been doing. Wish I'd paid more attention in high
school trig. But the girls just became way too interesting. Now I ]'m
old enough so that I can concentrate on the trig;-) Who knew it would
be useful -- the trig that is; the girls are obvious.

BP: Let x be position of an object relative to 0,0,0 in the left-right
direction, y be the position in the up-down direction, and z be the position
in the direction at right angles to the x and y axes.

b = ball, f = fielder:

Lateral angle = arctan{(x[b] - x[f])/(z[b] - z[f])}

Vertical angle = arctan{(y[b] - y[f])/(z[b] - z[f])}

distance, b to f: SQRT{sqr(x[b] - x[f]) + sqr(y[b] - y[f]) + sqr(z[b] -
z[f])}

Skip computing the arctan in the case where z[b] = z[f] where the angle
would be 90 degrees.

You may already be computing at least the lateral and vertical angles in the
model program.

RM: I compute it all; but I don't have a system that is controlling
for distance. Maybe I should!! I presume it can be perceived in terms
of the size of the object (not always a ball).

BP: You could also compute the first and second differences to look at the
velocity and acceleration of the angles. From the first differences of the
fielder's position you can get his direction and speed of movement.

RM: That's exactly what I'm doing. You are so 2010;-) But the distance
is something I never considered. I think I'll try it and see if it
fixes up the gain variance problem I told you about.

BP: You could show us the results of these computations without revealing the
data gathered by other people.

RM: Sure. I'll show you some results. Give me a minute-- or a couple days;-)

Best

Rick

路路路

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

[From bill Powers (2012.11.27.1740 MST)]

Rick Marken (2012.11.27.1625) --

RM: I compute it all; but I don't have a system that is controlling
for distance. Maybe I should!! I presume it can be perceived in terms
of the size of the object (not always a ball).

BP: I don't think you need to bother, unless you want to model the part about getting the glove to meet the ball after it gets close enough. I don't think that's the main point of the current modeling effort.

RM: That's exactly what I'm doing. You are so 2010;-) But the distance
is something I never considered. I think I'll try it and see if it
fixes up the gain variance problem I told you about.

BP: So 2010! Marken, sometimes you really break me up. I still laugh every time I read that.

The distance information would be interesting just to see how close the ball gets to where the fielder is.

Looking forward to the results.

Bill

[From Rick Marken (2012.11.27.1715)]

RM: Sure. I'll show you some results. Give me a minute-- or a couple days;-)

Well, it turns out it took only 1 minute (well, maybe 30) because I
have a figure handy that I think I can show you. It's attached. This
is data from a study of how people intercept toy helicopters. The data
shows 4 characteristic trials (there were about 50 trials). On each
trial a pursuer moves to intercept (catch) the helicopter. The data in
the figure show the four trials viewed from the top; you are seeing
movements over time in x,y coordinates of the helicopter (green line),
human pursuer (blue line) and the control model of the pursuer (red
line). The model is called COV for control of optical velocity. The
R2 values are a measure of fit of the model movements to the actual
human pursuer's movements. The fit is pretty good in terms of R2 --
always greater than .96. Fit was also measured as RMS deviation of
model from actual pursuer; the RMS deviation averaged about 16 mm; the
range of movement was over 4000 mm (4 meters for those of you still
living in non-metric countries). The helicopter moves in the z
dimension (up and down) but, of course, this is not shown in the
figure.

The model consists of two control systems: one controlling vertical
optical velocity (derived from the x,y,z, coordinates of the
helicopter in this case; it works with baseballs, cricket balls and
Frisbees too;-) by moving the model pursuer forward or back (in the y
-- vertical -- dimension) as necessary and the other controlling
lateral position -- basically keeping the optical movement of the
helicopter vertical -- by moving left or right, in the x dimension as
necessary.

The model works pretty well, I think. The model behaves almost exactly
like the actual pursuer on all trials. When I change one of the the
controlled perception from vertical velocity to vertical acceleration
the models does consistently worse, which is evidence that one of the
the controlled variable is closer to vertical velocity than to
vertical acceleration. Oh, and these perceptions are derived from
angular calculations, not tangents. It doesn't work well at all when
tangets are used, which makes sense since we perceive angles not
tangents.

Best

Rick

PursuitData.png

路路路

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

Content-Type: image/png; name="PursuitData.png"
Content-Disposition: attachment; filename="PursuitData.png"
X-Attachment-Id: f_ha1qc9rx0

[From Bill Powers (2012.11.28.0928 MST)]

Rick Marken (2012.11.27.1715) --

> RM: Sure. I'll show you some results. Give me a minute-- or a couple days;-)

Well, it turns out it took only 1 minute (well, maybe 30) because I
have a figure handy that I think I can show you. It's attached. This
is data from a study of how people intercept toy helicopters.

This isn't what I had in mind, though it does show how well the model reproduces the path of the real catcher. Unfortunately it doesn't show the error in the real positions -- that is, when the blue and red traces cross, it is unlikely that the real pursuer and the model pursuer are at exactly that intersection point at the same time, so the distance between them is probably not zero. I'm sure you did the R^2 calculations using samples representing the same times, but that doesn't show in the figure.

Because the figures you sent don't show corresponding catcher and object positions, we can't see in what direction the vector from catcher to object is pointing. The hypothesis is that the catcher moves so as to keep the object at a constant (or slowly changing) angle in three dimensions, which would come out to a constant azimuth and a constant elevation. I was hoping to see a plot of the angular direction from catcher to object at intervals during the run, in each dimension -- in other words, a vector from catcher to object. You could just plot the direction angle against time.

In Take 1, the pursuer (blue) starts directly north (+x) of the helicopter (green). Then both move to the left (+y, or west) while getting closer together in the x direction, but if one is moving faster or slower than the other along the indicated path the diagram doesn't show that. Just looking at all four plots, it looks as though the directions are maintained very roughly constant, but far from perfectly. The irregular path of the helicopter seems to be difficult for the pursuer to follow exactly. That's OK, but it doesn't handle the original case in which a ball moves smoothly in a regular curve, making control easier.

Also, the scale of this experiment is much smaller than the scale of a real ball park. The field is only about 16 feet square in one quadrant, or about 8 times one arm length of the pursuer. So it's hard to judge how far "the pursuer" is from the helicopter -- does the pursuer stop when he can reach out and grab the helicopter? There is a built-in uncertainty of the pursuer's position of about +/-60 centimeters, the distance from shoulder (or body centerline) to hand.

Do you have comparable data for the fly-ball case? I think this would be much more germane to the original question, even though it is less of a test of the control system dynamics because the task is a lot easier.

Best,

Bill P.

[From Rick Marken (2012.11.28.1515)]

Bill Powers (2012.11.28.0928 MST)]

RM Well, it turns out it took only 1 minute (well, maybe 30) because I
have a figure handy that I think I can show you. It's attached. This
is data from a study of how people intercept toy helicopters.

BP: This isn't what I had in mind,

RM: Talk about hard to please!!

BP: though it does show how well the model reproduces the path
of the real catcher.

RM: Don't it though;-)

BP: Unfortunately it doesn't show the error in the real positions ...
I'm sure you did the R^2 calculations using samples
representing the same times, but that doesn't show in the figure.

RM: I believe I did but there's really no telling with me;-)

BP: Because the figures you sent don't show corresponding catcher and object
positions, we can't see in what direction the vector from catcher to object
is pointing. The hypothesis is that the catcher moves so as to keep the
object at a constant (or slowly changing) angle in three dimensions, which
would come out to a constant azimuth and a constant elevation. I was hoping
to see a plot of the angular direction from catcher to object at intervals
during the run, in each dimension -- in other words, a vector from catcher
to object. You could just plot the direction angle against time.

That will take some time but I'll try to get it for you, using some
baseball catch data I have.

Best

Rick

路路路

In Take 1, the pursuer (blue) starts directly north (+x) of the helicopter
(green). Then both move to the left (+y, or west) while getting closer
together in the x direction, but if one is moving faster or slower than the
other along the indicated path the diagram doesn't show that. Just looking
at all four plots, it looks as though the directions are maintained very
roughly constant, but far from perfectly. The irregular path of the
helicopter seems to be difficult for the pursuer to follow exactly. That's
OK, but it doesn't handle the original case in which a ball moves smoothly
in a regular curve, making control easier.

Also, the scale of this experiment is much smaller than the scale of a real
ball park. The field is only about 16 feet square in one quadrant, or about
8 times one arm length of the pursuer. So it's hard to judge how far "the
pursuer" is from the helicopter -- does the pursuer stop when he can reach
out and grab the helicopter? There is a built-in uncertainty of the
pursuer's position of about +/-60 centimeters, the distance from shoulder
(or body centerline) to hand.

Do you have comparable data for the fly-ball case? I think this would be
much more germane to the original question, even though it is less of a test
of the control system dynamics because the task is a lot easier.

Best,

Bill P.

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