An (extreme?) example of Imagination Mode?

From Slashdot:

Soccer Superstar Plays With Very Low Brain Activity

timothy posted yesterday | from the one-hemisphere-tied-behind-his-back dept.

image0019.png136

jones_supa (887896) writes"Brazilian superstar Neymar’s (Neymar da Silva Santos Júnior) brain activity while dancing past opponents is less than 10 per cent the level of amateur players, suggesting he plays as if on “auto-pilot”, according to Japanese neurologists Eiichi Naito and Satoshi Hirose. The findings were published in the Swiss journal Frontiers in Human Neuroscience following a series of motor skills tests carried out on the 22-year-old Neymar and several other athletes in Barcelona in February this year. Three Spanish second-division footballers and two top-level swimmers were also subjected to the same tests. Researcher Naito told Japan’s Mainichi Shimbun newspaper: “Reduced brain activity means less burden which allows [the player] to perform many complex movements at once. We believe this gives him the ability to execute his various shimmies.” In the research paper Naito concluded that the test results “provide valuable evidence that the football brain of Neymar recruits very limited neural resources in the motor-cortical foot regions during foot movements”."

[From Rick Marken (2014.07.27.1225)]

image00110.png

···

On Sun, Jul 27, 2014 at 9:18 AM, Ted Cloak tcloak@unm.edu wrote:

From Slashdot:

Soccer Superstar Plays With Very Low Brain Activity

Hi Ted

It’s hard to know what to make of these neurophysiological studies, especially ones like this where they just measure gross neural activity. But it is fun to try to make up stories about what might be going on. My story about this article would be about reorganization. I think the overall level of neural activity could be an indication of the level of activity of the reorganizing system. Expert controllers, like Silva Santos, are controlling skillfully so the ambient level of error in their control hierarchy is very low while they are doing what they do best – in this case, play soccer. To the extent that a high ambient level of error in the control hierarchy increases the rate of reorganization – and to the extent that this reorganization process involves neural activity – the less skilled controllers (soccer players) would be expected to show higher levels of neural activity since they not only have higher levels of ambient error (which is definitely expected to be reflected in higher measured neural activity) but, therefore, they also have high rates of reorganization.

Whaddaya think?

Best

Rick

timothy posted yesterday | from the one-hemisphere-tied-behind-his-back dept.

136

jones_supa (887896) writes"Brazilian superstar Neymar’s (Neymar da Silva Santos Júnior) brain activity while dancing past opponents is less than 10 per cent the level of amateur players, suggesting he plays as if on “auto-pilot”, according to Japanese neurologists Eiichi Naito and Satoshi Hirose. The findings were published in the Swiss journal Frontiers in Human Neuroscience following a series of motor skills tests carried out on the 22-year-old Neymar and several other athletes in Barcelona in February this year. Three Spanish second-division footballers and two top-level swimmers were also subjected to the same tests. Researcher Naito told Japan’s Mainichi Shimbun newspaper: “Reduced brain activity means less burden which allows [the player] to perform many complex movements at once. We believe this gives him the ability to execute his various shimmies.” In the research paper Naito concluded that the test results “provide valuable evidence that the football brain of Neymar recruits very limited neural resources in the motor-cortical foot regions during foot movements”."


Richard S. Marken
Author of Doing Research on Purpose
Now available from Amazon or Barnes & Noble

image00110.png

···

from Slashdot:

[ Soccer

Superstar Plays With Very Low Brain Activity](http://science-beta.slashdot.org/story/14/07/26/138209/soccer-superstar-plays-with-very-low-brain-activity)

timothy posted

yesterday | from the one-hemisphere-tied-behind-his-back dept.

136

jones_supa (887896) writes"Brazilian
superstar Neymar 's (Neymar da Silva Santos
Júnior) brain activity while dancing past opponents is less
than 10 per cent the level of amateur players
, suggesting he plays as if on “auto-pilot”, according to Japanese neurologists Eiichi Naito and Satoshi Hirose. The findings were published in the Swiss journal Frontiers in Human Neuroscience following
a series of motor skills tests carried out on the 22-year-old Neymar and several other athletes in Barcelona in February this year. Three Spanish second-division footballers and two top-level swimmers were also subjected to the same tests. Researcher Naito
told Japan’s Mainichi Shimbun newspaper: “Reduced brain activity means less burden which allows [the player] to perform many complex movements at once. We believe this gives him the ability to execute his various shimmies.” In
the research paper Naito concluded
that the test results “provide valuable evidence that the football brain of Neymar recruits very limited neural resources in the motor-cortical foot regions during foot
movements”."

image00112.png

···

from Slashdot:

[ Soccer

Superstar Plays With Very Low Brain Activity](http://science-beta.slashdot.org/story/14/07/26/138209/soccer-superstar-plays-with-very-low-brain-activity)

timothy posted

yesterday | from the one-hemisphere-tied-behind-his-back dept.

136

jones_supa (887896) writes"Brazilian
superstar Neymar 's (Neymar da Silva Santos
Júnior) brain activity while dancing past opponents is less
than 10 per cent the level of amateur players
, suggesting he plays as if on “auto-pilot”, according to Japanese neurologists Eiichi Naito and Satoshi Hirose. The findings were published in the Swiss journal Frontiers in Human Neuroscience following
a series of motor skills tests carried out on the 22-year-old Neymar and several other athletes in Barcelona in February this year. Three Spanish second-division footballers and two top-level swimmers were also subjected to the same tests. Researcher Naito
told Japan’s Mainichi Shimbun newspaper: “Reduced brain activity means less burden which allows [the player] to perform many complex movements at once. We believe this gives him the ability to execute his various shimmies.” In
the research paper Naito concluded
that the test results “provide valuable evidence that the football brain of Neymar recruits very limited neural resources in the motor-cortical foot regions during foot
movements”."

[Martin Taylor 2104.07.27.17.40]

[From Kent McClelland (2014.07.27.1520 EDT)]

Hi Ted,

    It seems to me to be a reasonable hypothesis that, in the

kind of reorganization of control systems that produces highly
skilled and automatic motor actions, the relevant perceptions,
in effect, move down in hierarchical levels.

Don't you mean "up" in the levels? Each level up coalesces several

lower-level perceptions into a unitary perception, doesn’t it? And
when the higher level sets references for the lower levels, and the
lower levels are acting skilfully, the higher-level perception only
has to deal with disturbances that are intrinsically at that level.
In the soccer game, that might mean that a tactical situation
involving several players on each side is perceived just as a unit
rather than as “this player and where is he going” and “where is the
ball going to be” and “why is that player moving where I would not
have expected”. And as a unit, the error in the tactical situation
perception can be converted into reference values for the
lower-level systems, ending in muscular output – “I’m going
there at this speed”.

    Chunks of physical action that were perceived as sequences of

relationships, events, and transitions, for instance, become
unitary actions that can be perceived (and executed) at the
configuration level. The resulting perceptions at the
configuration level can then be controlled much more quickly and
with little mental effort.

Yes, in principle, though I’m not going to name levels.

Martin

from Fred Nickols 7/27/2014 6:45 pm

Martin:

Or, as Wayne Gretzky famously said, "I skate to where the puck will be."

In PCT terms, I think he was getting at a reference state that had something to do with where he thought he ought to be in light of where he thought the puck was going to be.

Fred Nickols
Managing Partner
Distance Consulting LLC
Be sure you measure what you want.
Be sure you want what you measure.

···

Sent from my iPad

On Jul 27, 2014, at 5:48 PM, Martin Taylor <mmt-csg@mmtaylor.net> wrote:

[Martin Taylor 2104.07.27.17.40]

[From Kent McClelland (2014.07.27.1520 EDT)]

Hi Ted,

It seems to me to be a reasonable hypothesis that, in the kind of reorganization of control systems that produces highly skilled and automatic motor actions, the relevant perceptions, in effect, move down in hierarchical levels.

Don't you mean "up" in the levels? Each level up coalesces several lower-level perceptions into a unitary perception, doesn't it? And when the higher level sets references for the lower levels, and the lower levels are acting skilfully, the higher-level perception only has to deal with disturbances that are intrinsically at that level. In the soccer game, that might mean that a tactical situation involving several players on each side is perceived just as a unit rather than as "this player and where is he going" and "where is the ball going to be" and "why is that player moving where I would not have expected". And as a unit, the error in the tactical situation perception can be converted into reference values for the lower-level systems, ending in muscular output -- "I'm going _there_ at _this_ speed".

Chunks of physical action that were perceived as sequences of relationships, events, and transitions, for instance, become unitary actions that can be perceived (and executed) at the configuration level. The resulting perceptions at the configuration level can then be controlled much more quickly and with little mental effort.

Yes, in principle, though I'm not going to name levels.

Martin

Content-Type: application/pdf; name="RobertsonGlinesPaper.pdf"
Content-Disposition: attachment; filename="RobertsonGlinesPaper.pdf"
X-Attachment-Id: f_hy4zu2gi0

Content-Type: application/pdf; name="PowersCommentonPlateau.pdf"
Content-Disposition: attachment; filename="PowersCommentonPlateau.pdf"
X-Attachment-Id: f_hy4zv3ab1

RobertsonGlinesPaper.pdf (310 KB)

PowersCommentonPlateau.pdf (58.9 KB)

[From Kent McClelland (2014.10.28.1100)]

Hello Rick and Martin,

KM: Yes, of course, I take your point. The “chunking” (as Robertson and Glines put it) of motor actions and perceptions that happens with reorganizations means that lots of lower-order perceptions get consolidated into a single higher-order perceptual
unit that can be controlled without any need for attention to the automatically controlled details of the lower-order perceptions contributing to it.

KM: What interests me about this phenomenon, however, is the question of the “reaction time” (again using the terms from Robertson and Glines) necessary for controlling this reorganized higher-order perception. It’s a general rule with a perceptual hierarchy
that perception processing occurs more rapidly as you go down in levels, or otherwise hierarchical control would not be stable.

KM: But it appears that these reorganized higher-order perceptions can be controlled very rapidly, or at least that’s what the Robertson and Glines data seems to show if I’m reading it correctly. The chunked higher-order action sequences are executed even
more quickly than the individual components of the sequence had been before the reorganization. It’s as if the higher-order perception had moved down in levels in terms of the speed of its control.

KM: When a higher-order perception has been consolidated as a unit, we often start giving it a name or a label, which is a lower-order perception that symbolizes this higher-order complex, if I’m thinking about it correctly. A skilled chess player, for
instance, has only to glance a chess board and see that the opening was king’s gambit, with all the strengths and weaknesses inherent in that board position. Or a skilled musician can hear a series of notes and think, “arpeggio in D minor,” and then instantly
play it on his own instrument. (I’m a little out of my depth here, being neither a skilled chess player nor musician, but I presume you get my point.)

KM: Some of the increase in speed must come from the “optimizing of parameters” of control, as Rick puts it, but this optimized perceptual chunk also seems to be controlled as if it were a lower-level perception in terms of speed. What am I missing?

By the way, thanks Rick, for distributing the Robertson and Glines paper along with Bill’s reply. It’s very interesting, and I don’t remember having read it before.

Best to all,

Kent

···

[From Rick Marken (2014.07.27.1650)]

Martin Taylor (2104.07.27.17.40)–

Kent McClelland (2014.07.27.1520 EDT)–

KM: It seems to me to be a reasonable hypothesis that, in the kind of reorganization of control systems that produces highly skilled and automatic motor actions, the relevant perceptions, in effect, move down in hierarchical levels.

MT: Don’t you mean “up” in the levels? Each level up coalesces several lower-level perceptions into a unitary perception, doesn’t it?

RM: You beat me to the punch! I was going to make exactly the same point. PCT suggests that increased skill results from not only from optimizing the parameters of all control systems involved in the behavior but also from controlling higher level perceptions
that make up the skill. The skilled soccer player is not only controlling the variables involved in stopping and kicking the ball but also the “field position” variables that are the reason why the ball is being stopped (or not) and kicked to a particular
player.

[From Rick Marken (2014.07.29.1800)]

···

Kent McClelland (2014.10.28.1100)–

KM: Yes, of course, I take your point.

KM: What interests me about this phenomenon, however, is the question of the “reaction time” (again using the terms from Robertson and Glines) necessary for controlling this reorganized higher-order perception. It’s a general rule with a perceptual hierarchy
that perception processing occurs more rapidly as you go down in levels, or otherwise hierarchical control would not be stable.

KM: But it appears that these reorganized higher-order perceptions can be controlled very rapidly, or at least that’s what the Robertson and Glines data seems to show if I’m reading it correctly.

RM: This is an excellent point. Higher level control systems – control systems that control complex perceptions, like sequences, are presumed to have a longer “reaction time” (which in a control system involves both transport lag and integration of output over time) than lower level ones – which control simpler perceptions, like intensities and shapes. And there is evidence that this is the case (eg. Marken, R. S., Khatib, Z. and Mansell, W. (2013) Motor Control as the
Control of Perception, Perceptual and
Motor Skills
, 117, 236-247).

RM: But the measures of reaction time in the Robertson/Glines are not pure measures of the speed of control at different levels of perception. As noted in the paper, the reaction times are somewhat constrained by the nature of the task itself. Being able to react correctly and quickly – indeed, eventually anticipating the correct response – is what defined mastery in this task. You master the task by learning to control a sequence which allows you to press the appropriate key even before the associated display element occurs. So when there is perfect mastery of the task, reaction time – which is measured as the time between display item and keypress – is actually negative. Reaction time in this task is, therefore, definitely not a measure of the temporal characteristics of control. Reaction time is actually itself a controlled variable; you have to learn keep reaction time negative – learn to anticipate the next item in the sequence – in order to win the game.

RM: But it is possibly confusing to someone who understands hierarchical control to use reaction time as a measure of learning to control progressively higher level perceptions, expecting reaction time to be lowest when the participant is controlling the highest level variable. Maybe someone could design a version of this study that didn’t use reaction time as the measure performance.I can’t think of anything offhand but maybe some of the clever people out there can.

Best

Rick


Richard S. Marken
Author of Doing Research on Purpose
Now available from Amazon or Barnes & Noble

KM: It seems to me to be a reasonable hypothesis that, in the kind of reorganization of control systems that produces highly skilled and automatic motor actions, the relevant perceptions, in effect, move down in hierarchical levels.

MT: Don’t you mean “up” in the levels? Each level up coalesces several lower-level perceptions into a unitary perception, doesn’t it?

RM: You beat me to the punch! I was going to make exactly the same point.

Hi guys,

KM: It seems to me to be a reasonable hypothesis that, in the kind of reorganization of control systems that produces highly skilled and automatic motor actions, the relevant perceptions, in effect, move down in hierarchical levels.

HB :

I’m inclined to beleive that Kent is right as far the »reaction times«, smoothness, precisiion of movings and other skills of Live organisms are in question. But I must admitt that I was never thinking about this problem before in PCT dimensions. I don’t know how could it work in control systems as Kent used that term. But I could agree what he said about highly skilled and automatiozation of movements, which are produced by lower order parts of nervous system in highly skilled players.

So maybe it could not be bad if we divide for a moment »Live organism« and »PCT-organism« and try to understand why there is a difference in explanations. I’m loooking form the point of very skilled sporstman as I was when I was young. It seems that in the process of training (couple hours a day) repetitions lead to highly skilled and automatic motor actions (controlled perceptions), which seems to be moving down the levels of nervous system, which take over control of »simple movements«. It looks like skilled player is not aware of most body movements.

For exmple table tennis : begginer has to think mostly of his body movements, controling mostly balance perceptions, perception of hands movement, perception of the flight of the ball in coordination of hand movement and so on. While the skilled player controlls mostly perception of other player’s movements, effects which ball have, anticipating where opponent will place the ball and so on. Skilled player is practically not aware of coordinated body, legs or hands movement. He seems not to think about the problems begginer has.

So it seems that at first higher order perceptions in the begginig of training are intensities of coordinated actions of body parts and later probably conjtrol proceeds through other levels of control to levels for controlling relations to ball and system concepts like relations to other player’s position and wholw playing situation. Skiled player seems to control for very complex percpetions of spots on the table where he will place the ball, how much acceleration and spin will he »put« in the ball in relation to position of other player. I have a feeling that there is also a lot of anticipations (probably a lot of controlling in and off imagination).

It was just a flash. But I think it’s not so easy to define this problem with PCT means.

Best,

Boris

···

From: csgnet-request@lists.illinois.edu [mailto:csgnet-request@lists.illinois.edu] On Behalf Of Richard Marken
Sent: Wednesday, July 30, 2014 3:01 AM
To: csgnet@lists.illinois.edu
Subject: Re: An (extreme?) example of Imagination Mode?

[From Rick Marken (2014.07.29.1800)]

Kent McClelland (2014.10.28.1100)–

KM: It seems to me to be a reasonable hypothesis that, in the kind of reorganization of control systems that produces highly skilled and automatic motor actions, the relevant perceptions, in effect, move down in hierarchical levels.

MT: Don’t you mean “up” in the levels? Each level up coalesces several lower-level perceptions into a unitary perception, doesn’t it?

RM: You beat me to the punch! I was going to make exactly the same point.

KM: Yes, of course, I take your point.

KM: What interests me about this phenomenon, however, is the question of the “reaction time” (again using the terms from Robertson and Glines) necessary for controlling this reorganized higher-order perception. It’s a general rule with a perceptual hierarchy that perception processing occurs more rapidly as you go down in levels, or otherwise hierarchical control would not be stable.

KM: But it appears that these reorganized higher-order perceptions can be controlled very rapidly, or at least that’s what the Robertson and Glines data seems to show if I’m reading it correctly.

RM: This is an excellent point. Higher level control systems – control systems that control complex perceptions, like sequences, are presumed to have a longer “reaction time” (which in a control system involves both transport lag and integration of output over time) than lower level ones – which control simpler perceptions, like intensities and shapes. And there is evidence that this is the case (eg. Marken, R. S., Khatib, Z. and Mansell, W. (2013) Motor Control as the Control of Perception, Perceptual and Motor Skills, 117, 236-247).

RM: But the measures of reaction time in the Robertson/Glines are not pure measures of the speed of control at different levels of perception. As noted in the paper, the reaction times are somewhat constrained by the nature of the task itself. Being able to react correctly and quickly – indeed, eventually anticipating the correct response – is what defined mastery in this task. You master the task by learning to control a sequence which allows you to press the appropriate key even before the associated display element occurs. So when there is perfect mastery of the task, reaction time – which is measured as the time between display item and keypress – is actually negative. Reaction time in this task is, therefore, definitely not a measure of the temporal characteristics of control. Reaction time is actually itself a controlled variable; you have to learn keep reaction time negative – learn to anticipate the next item in the sequence – in order to win the game.

RM: But it is possibly confusing to someone who understands hierarchical control to use reaction time as a measure of learning to control progressively higher level perceptions, expecting reaction time to be lowest when the participant is controlling the highest level variable. Maybe someone could design a version of this study that didn’t use reaction time as the measure performance.I can’t think of anything offhand but maybe some of the clever people out there can.

Best

Rick

Richard S. Marken
Author of Doing Research on Purpose

Now available from Amazon or Barnes & Noble

[Martin Taylor 2014.07.30.10.51]

Boris,

What do you think is different between what you say in your last two
and a half big paragraphs and what Rick and i said in our comments?
You seem to be giving examples of exactly what we said in a more
general way. And why do you think it is difficult to define it in
PCT terms, when that is exactly what Kent, Rick, and I all did?
Just as a self-advertisement, it is also what Insup and I described
30 years ago (though not in PCT terms) as happening when one learns
to read fluently (I. Taylor and M. Taylor, The Psychology of
Reading, Academic Press, 1983). We called it “Three-phase learing”,
where one learns the constituent parts (e.g. letters) as entities,
then puts them together individually, and finally sees different
groupings as entities in their own right (e.g. words). You, Rick, I, and our old book all see what is happening as the
combination of repeated patterns into simple higher-level
perceptions. It might even be true!
Martin

···

On 2014/07/30 10:44 AM, Boris Hartman
wrote:

        Hi

guys,

      KM: It seems to me to be a reasonable

hypothesis that, in the kind of reorganization of control
systems that produces highly skilled and automatic motor
actions, the relevant perceptions, in effect, move down in
hierarchical levels.

        HB

:

        I'm

inclined to beleive that Kent is right as far the »reaction
times«, smoothness, precisiion of movings and other skills
of Live organisms are in question. But I must admitt that I
was never thinking about this problem before in PCT
dimensions. I don’t know how could it work in control
systems as Kent used that term. But I could agree what he
said about highly skilled and automatiozation of movements,
which are produced by lower order parts of nervous system in
highly skilled players.

        So

maybe it could not be bad if we divide for a moment »Live
organism« and »PCT-organism« and try to understand why there
is a difference in explanations. I’m loooking form the point
of very skilled sporstman as I was when I was young. It
seems that in the process of training (couple hours a day)
repetitions lead to highly skilled and automatic motor
actions (controlled perceptions), which seems to be moving
down the levels of nervous system, which take over control
of »simple movements«. It looks like skilled player is not
aware of most body movements.

        For

exmple table tennis : begginer has to think mostly of his
body movements, controling mostly balance perceptions,
perception of hands movement, perception of the flight of
the ball in coordination of hand movement and so on. While
the skilled player controlls mostly perception of other
player’s movements, effects which ball have, anticipating
where opponent will place the ball and so on. Skilled player
is practically not aware of coordinated body, legs or hands
movement. He seems not to think about the problems begginer
has.

        So

it seems that at first higher order perceptions in the
begginig of training are intensities of coordinated actions
of body parts and later probably conjtrol proceeds through
other levels of control to levels for controlling relations
to ball and system concepts like relations to other player’s
position and wholw playing situation. Skiled player seems to
control for very complex percpetions of spots on the table
where he will place the ball, how much acceleration and spin
will he »put« in the ball in relation to position of other
player. I have a feeling that there is also a lot of
anticipations (probably a lot of controlling in and off
imagination).

        It

was just a flash. But I think it’s not so easy to define
this problem with PCT means.

Best,

Boris

Martin,

Well I didn’t read all carefully. It seemed to me that there could be some conflict between you and Kent using »up« and »down« terms. Sorry for misunderstanding J

Boris

···

From: csgnet-request@lists.illinois.edu [mailto:csgnet-request@lists.illinois.edu] On Behalf Of Martin Taylor
Sent: Wednesday, July 30, 2014 5:01 PM
To: csgnet@lists.illinois.edu
Subject: Re: An (extreme?) example of Imagination Mode?

[Martin Taylor 2014.07.30.10.51]

On 2014/07/30 10:44 AM, Boris Hartman wrote:

Hi guys,

KM: It seems to me to be a reasonable hypothesis that, in the kind of reorganization of control systems that produces highly skilled and automatic motor actions, the relevant perceptions, in effect, move down in hierarchical levels.

HB :

I’m inclined to beleive that Kent is right as far the »reaction times«, smoothness, precisiion of movings and other skills of Live organisms are in question. But I must admitt that I was never thinking about this problem before in PCT dimensions. I don’t know how could it work in control systems as Kent used that term. But I could agree what he said about highly skilled and automatiozation of movements, which are produced by lower order parts of nervous system in highly skilled players.

So maybe it could not be bad if we divide for a moment »Live organism« and »PCT-organism« and try to understand why there is a difference in explanations. I’m loooking form the point of very skilled sporstman as I was when I was young. It seems that in the process of training (couple hours a day) repetitions lead to highly skilled and automatic motor actions (controlled perceptions), which seems to be moving down the levels of nervous system, which take over control of »simple movements«. It looks like skilled player is not aware of most body movements.

For exmple table tennis : begginer has to think mostly of his body movements, controling mostly balance perceptions, perception of hands movement, perception of the flight of the ball in coordination of hand movement and so on. While the skilled player controlls mostly perception of other player’s movements, effects which ball have, anticipating where opponent will place the ball and so on. Skilled player is practically not aware of coordinated body, legs or hands movement. He seems not to think about the problems begginer has.

So it seems that at first higher order perceptions in the begginig of training are intensities of coordinated actions of body parts and later probably conjtrol proceeds through other levels of control to levels for controlling relations to ball and system concepts like relations to other player’s position and wholw playing situation. Skiled player seems to control for very complex percpetions of spots on the table where he will place the ball, how much acceleration and spin will he »put« in the ball in relation to position of other player. I have a feeling that there is also a lot of anticipations (probably a lot of controlling in and off imagination).

It was just a flash. But I think it’s not so easy to define this problem with PCT means.

Best,

Boris

Boris,

What do you think is different between what you say in your last two and a half big paragraphs and what Rick and i said in our comments? You seem to be giving examples of exactly what we said in a more general way. And why do you think it is difficult to define it in PCT terms, when that is exactly what Kent, Rick, and I all did?

Just as a self-advertisement, it is also what Insup and I described 30 years ago (though not in PCT terms) as happening when one learns to read fluently (I. Taylor and M. Taylor, The Psychology of Reading, Academic Press, 1983). We called it “Three-phase learing”, where one learns the constituent parts (e.g. letters) as entities, then puts them together individually, and finally sees different groupings as entities in their own right (e.g. words).

You, Rick, I, and our old book all see what is happening as the combination of repeated patterns into simple higher-level perceptions. It might even be true!

Martin

[Martin Taylor 2104.07.30.13.45]

Yes, Rick and I did disagree with Kent on "up" and "down", but we

all treated the problem in PCT terms, which is what you suggested
would be difficult to do.
Unless I misread Kent’s original message, he seemed actually to be
saying that the individual elements of the skilled pattern had
coalesced to form a controllable perception. []
Rick and I disagreed with him about where this unitary perception
would be – up or down from its constituent part perceptions.
The problem I understood him to be presenting was how that could
lead to faster apparent reaction time, when the usual statement is
that higher levels control more slowly than lower levels. What he
pushed “down” was to increase speed by using only the lower levels
– a puzzle, considering that they had already reorganized well. Rick provided an answer to that puzzle. Maybe it was a correct
answer, maybe it wasn’t, but I think something along Rick’s lines
makes a lot of sense.
Martin

···

On 2014/07/30 1:31 PM, Boris Hartman
wrote:

Martin,

        Well

I didn’t read all carefully. It seemed to me that there
could be some conflict between you and Kent using »up« and
»down« terms. Sorry for misunderstanding J

Boris

  •  KM: Chunks of
    

physical action that were perceived as sequences of relationships,
events, and transitions, for instance, become unitary actions that
can be perceived (and executed) at the configuration level.*

[From Kent McClelland (2014.07.30.1405)]

Martin Taylor 2104.07.30.13.45

KM: Martin’s post (below) is a fair statement of the “puzzle” I was presenting (as he puts it), the disagreement we had, and the possible solution Rick offered.

Perhaps I’ve been reading too many non-PCT discussions of brain architecture, because I’ve seen this kind of claim before—that complex sequences of action get “chunked” into packages that are then “executed” as if lower in a hierarchical arrangement.
The citation that sticks in my mind to this effect is Jeff Hawkins’ book, ON INTELLIGENCE (2004).

Hawkins presents a theory that he thinks is blindingly original (without any knowledge, apparently, of Powers’ work), which includes a hierarchical architecture for the brain similar in some respects to HPCT. Instead of control system units, however, he
proposes that the brain sends down “predictions” from higher levels that are then confirmed or disconfirmed (he doesn’t specify exactly how) by incoming perceptions. I think there’s been some discussion of his book previously on CSGnet, though I don’t remember
exactly when.

In any case, the idea that actions get chunked and can then be more speedily executed is something that PCT advocates should be able to explain convincingly in PCT terms, if we want to converse with the wider world of cognitive sciences.

Kent

···

On 2014/07/30 1:31 PM, Boris Hartman wrote:

Martin,

Well I didn’t read all carefully. It seemed to me that there could be some conflict between you and Kent using »up« and »down« terms. Sorry for misunderstanding
J

Boris

  • KM: Chunks of physical action that were perceived as sequences of relationships,
    events, and transitions, for instance, become unitary actions that can be perceived (and executed) at the configuration level.*

[From Rick Marken (2014.07.30.1240)]

···

Kent McClelland (2014.07.30.1405)–

Martin Taylor 2104.07.30.13.45

KM: Martin’s post (below) is a fair statement of the “puzzle” I was presenting (as he puts it), the disagreement we had, and the possible solution Rick offered.

KM: Perhaps I’ve been reading too many non-PCT discussions of brain architecture, because I’ve seen this kind of claim before—thaat complex sequences of action get “chunked” into packages that are then “executed” as if lower in a hierarchical arrangement.
The citation that sticks in my mind to this effect is Jeff Hawkins’ book, ON INTELLIGENCE (2004).Â

RM: That idea has been around for a long time in psychology. And it’s basically correct – the chunking part anyway. But, of course, what is chunked, according to PCT, is input (perception), not output (muscle forces).Â

Â

KM: Hawkins presents a theory that he thinks is blindingly original (without any knowledge, apparently, of Powers’ work), which includes a hierarchical architecture for the brain similar in some respects to HPCT.

RM: There was actually a hierarchical control of output model of behavior in the same issues of Byte in which Bill published the “Nature of Robots” paper on hierarchical PCT. Plus ça change…

KM: In any case, the idea that actions get chunked and can then be more speedily executed is something that PCT advocates should be able to explain convincingly in PCT terms, if we want to converse with the wider world of cognitive sciences.

RM: Completely agree!! I will try to think of a demo that shows this. Actually, it may be as simple as showing that one can track a sinusoidal target moving at a much faster rate than a randomly moving one. With the random target you can only control the distance between target and cursor, a configuration; with the sine wave target you can control the sinusoidal sequence of configurations (target/cursor distances). The chunking is of the movements of the configuration into a perception of sinusoidal movement. This is seems like a good example of a case where the chunking can’t possibly be on the output side since the same output is being used to in the sine wave and random target case.Â

Best regards

Rick

Â

Kent

On Jul 30, 2014, at 1:56 PM, Martin Taylor wrote:

On 2014/07/30 1:31 PM, Boris Hartman wrote:

Martin,

Â

Well I didn’t read all carefully. It seemed to me that there could be some conflict between you and Kent using »up« and »down« terms. Sorry for misunderstanding
J

Â

Boris Â

Yes, Rick and I did disagree with Kent on “up” and “down”, but we all treated the problem in PCT terms, which is what you suggested would be difficult to do.
Unless I misread Kent’s original message, he seemed actually to be saying that the individual elements of the skilled pattern had coalesced to form a controllable perception. [* KM: Chunks of physical action that were perceived as sequences of relationships,
events, and transitions, for instance, become unitary actions that can be perceived (and executed) at the configuration level.*] Rick and I disagreed with him about where this unitary perception would be – up or down from its constituent part perceptions.

The problem I understood him to be presenting was how that could lead to faster apparent reaction time, when the usual statement is that higher levels control more slowly than lower levels. What he pushed “down” was to increase speed by using only the lower
levels – a puzzle, considering that they had already reorganized well.

Rick provided an answer to that puzzle. Maybe it was a correct answer, maybe it wasn’t, but I think something along Rick’s lines makes a lot of sense.

Martin


Richard S. Marken
Author of Doing Research on Purpose
Now available from Amazon or Barnes & Noble