FAST FORWARD Scientists pinpoint how brain tracks speeding objects

[From today’s Albuquerque Journal, p. C4. I have the Neuron article if anyone’s interested – Ted]

FAST FORWARD

Scientists pinpoint how brain tracks speeding objects

** ** By Lisa M. Krieger

** ** McClatchy Newspapers

 SAN JOSE, Calif. — The human brain is far slower than a Major League fastball or a blistering tennis serve — but it has figured out a workaround.

New research by University of California-Berkeley scientists solves a puzzle that has long mystified anyone who has watched, in awe, as elite athletes respond to incoming balls that can surpass 90 mph.

The brain perceives speeding objects as further along in their trajectory than seen by the eyes, giving us time to respond, according to research by Gerrit Maus, lead author of a paper published in a recent issue of the journal Neuron.

This clever adjustment — compensating for the sluggish route from the eyes to neural decision-making — “is a sophisticated prediction mechanism,” he said.

“As soon as the brain knows something is moving, it pushes the position of the object moving forward, so there’s a more accurate measure of where this object actually is,” said Maus.

This is useful in survival situations far more important than sports — such as when we’re crossing a street, in front of a speeding car.

Former Yankees catcher Yogi Berra pondered the mystery, once asking: “How can you think and hit at the same time?”

You can’t, because there’s not time for both.

“But you don’t need to think about it, because the brain does it automatically,” said Maus.

At the average major league speed of 90 mph, a baseball leaves the pitcher’s hand and travels about 56 feet to home plate in only 0.4 seconds, or 400 milliseconds.

Tennis is even faster. Last May, courtside radar guns measured a serve by British player Samuel Groth at 163 mph.

In that split second, there’s a lot of work for the body to do. Eyes must first find the ball. The sensory cells in the retina determine its speed and rush this information to the brain. Then the brain sends messages through the spinal cord that tell muscles in the arms and legs to respond.

“By time the brain receives the information, it’s already out of date,” said Maus.

The researchers said it can take one-tenth of a second for the brain to process what the eye sees. That means, for example, that by the time the brain “catches up” with incoming visual information, a fast moving tennis or baseball would already have moved 10 to 15 feet closer than the image in the eye.

A region in the back of the brain, called area V5, computes information about motion and position — and projects where it thinks the ball should be, rather than where the eyes saw it.

For the experiment, six volunteers had their brains scanned with a functional MRI as they viewed the “flash-drag effect,” a two-part visual illusion in which we see brief flashes shifting in the direction of a motion.

The researchers found that the illusion — flashes perceived in their predicted locations against a moving background and flashes actually shown in their predicted location against a still background — created the same neural activity patterns in the V5 region of the brain.

In an earlier study, they stimulated this part of the brain to interfere with neural activity, and disrupted this visual position-shifting mechanism.

The finding could also help explain why altered trajectories can fool us — such as tennis backspins or baseball pitches with so-called late break.

A clearer understanding of how the brain processes objects in motion can eventually help in diagnosing and treating myriad disorders, including those that impair motion perception, according to the UC Berkeley team. People who cannot perceive motion cannot predict locations of objects and therefore cannot perform tasks as simple as pouring a cup of coffee or crossing a road, researchers said.

“The brain doesn’t work in real time,” said Maus.

···

Science thrives on unanswered questions.

Religion thrives on unquestioned answers.

[From Rick Marken (2013.05.27.1015)]

[From today’s Albuquerque Journal, p. C4. I have the Neuron article if anyone’s interested – Ted]

FAST FORWARD

Scientists pinpoint how brain tracks speeding objects

** ** By Lisa M. Krieger

** ** McClatchy Newspapers

 SAN JOSE, Calif. — The human brain is far slower than a Major League fastball or a blistering tennis serve — but it has figured out a workaround.

New research by University of California-Berkeley scientists solves a puzzle that has long mystified anyone who has watched, in awe, as elite athletes respond to incoming balls that can surpass 90 mph.

The brain perceives speeding objects as further along in their trajectory than seen by the eyes, giving us time to respond, according to research by Gerrit Maus, lead author of a paper published in a recent issue of the journal Neuron.

Thanks, Ted. I can’t think of a more fitting example of why PCT is not better known and understood. The “feedback too slow” myth lives on.

Best

Rick

···

On Mon, May 27, 2013 at 10:02 AM, Ted Cloak tcloak@unm.edu wrote:

This clever adjustment — compensating for the sluggish route from the eyes to neural decision-making — “is a sophisticated prediction mechanism,” he said.

“As soon as the brain knows something is moving, it pushes the position of the object moving forward, so there’s a more accurate measure of where this object actually is,” said Maus.

This is useful in survival situations far more important than sports — such as when we’re crossing a street, in front of a speeding car.

Former Yankees catcher Yogi Berra pondered the mystery, once asking: “How can you think and hit at the same time?”

You can’t, because there’s not time for both.

“But you don’t need to think about it, because the brain does it automatically,” said Maus.

At the average major league speed of 90 mph, a baseball leaves the pitcher’s hand and travels about 56 feet to home plate in only 0.4 seconds, or 400 milliseconds.

Tennis is even faster. Last May, courtside radar guns measured a serve by British player Samuel Groth at 163 mph.

In that split second, there’s a lot of work for the body to do. Eyes must first find the ball. The sensory cells in the retina determine its speed and rush this information to the brain. Then the brain sends messages through the spinal cord that tell muscles in the arms and legs to respond.

“By time the brain receives the information, it’s already out of date,” said Maus.

The researchers said it can take one-tenth of a second for the brain to process what the eye sees. That means, for example, that by the time the brain “catches up” with incoming visual information, a fast moving tennis or baseball would already have moved 10 to 15 feet closer than the image in the eye.

A region in the back of the brain, called area V5, computes information about motion and position — and projects where it thinks the ball should be, rather than where the eyes saw it.

For the experiment, six volunteers had their brains scanned with a functional MRI as they viewed the “flash-drag effect,” a two-part visual illusion in which we see brief flashes shifting in the direction of a motion.

The researchers found that the illusion — flashes perceived in their predicted locations against a moving background and flashes actually shown in their predicted location against a still background — created the same neural activity patterns in the V5 region of the brain.

In an earlier study, they stimulated this part of the brain to interfere with neural activity, and disrupted this visual position-shifting mechanism.

The finding could also help explain why altered trajectories can fool us — such as tennis backspins or baseball pitches with so-called late break.

A clearer understanding of how the brain processes objects in motion can eventually help in diagnosing and treating myriad disorders, including those that impair motion perception, according to the UC Berkeley team. People who cannot perceive motion cannot predict locations of objects and therefore cannot perform tasks as simple as pouring a cup of coffee or crossing a road, researchers said.

“The brain doesn’t work in real time,” said Maus.


Science thrives on unanswered questions.

Religion thrives on unquestioned answers.


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

[Martin Taylor 2013.05.27.14.49]

Where do get this from the cited text? That they say it is

mysterious how it is possible for skilled athletes to perform as
they do? All I can see in the text is that they demonstrated that
the visual system can and sometimes does report a predicted location
of a moving object as though it was the actual position. Whether the
experiment suffices to demonstrate this is something I cannot judge
from the new report, but neither can I see in the news report
anything that suggests “feedback is too slow”. Methinks you are perceiving what you predict to be in the text.
Martin

···

[From Rick Marken (2013.05.27.1015)]

    On Mon, May 27, 2013 at 10:02 AM, Ted

Cloak tcloak@unm.edu
wrote:

              [From

today’s Albuquerque Journal, p. C4. I have the Neuron
article if anyone’s interested – Ted]

              FAST

FORWARD

Scientists pinpoint how brain tracks speeding objects

** **** By
Lisa M. Krieger**

** ** ** McClatchy
Newspapers**

              SAN JOSE, Calif. — The human brain is far slower than

a Major League fastball or a blistering tennis serve —
but it has figured out a workaround.

              New research by University of California-Berkeley

scientists solves a puzzle that has long mystified
anyone who has watched, in awe, as elite athletes
respond to incoming balls that can surpass 90 mph.

              The brain perceives speeding objects as further along

in their trajectory than seen by the eyes, giving us
time to respond, according to research by Gerrit Maus,
lead author of a paper published in a recent issue of
the journal Neuron.

      Thanks, Ted. I can't think of a more fitting example of why

PCT is not better known and understood. The “feedback too
slow” myth lives on.

              This clever adjustment — compensating for the sluggish

route from the eyes to neural decision-making — “is a
sophisticated prediction mechanism,” he said.

              “As soon as the brain knows something is moving, it

pushes the position of the object moving forward, so
there’s a more accurate measure of where this object
actually is,” said Maus.

              This is useful in survival situations far more

important than sports — such as when we’re crossing a
street, in front of a speeding car.

              Former Yankees catcher Yogi Berra pondered the

mystery, once asking: “How can you think and hit at
the same time?”

              You can’t, because there’s not time for both.
              “But you don’t need to think about it, because the

brain does it automatically,” said Maus.

              At the average major   league speed of 90 mph, a

baseball leaves the pitcher’s hand and travels about
56 feet to home plate in only 0.4 seconds, or 400
milliseconds.

              Tennis is even faster. Last May, courtside radar guns

measured a serve by British player Samuel Groth at 163
mph.

              In that split second, there’s a lot of work for the

body to do. Eyes must first find the ball. The sensory
cells in the retina determine its speed and rush this
information to the brain. Then the brain sends
messages through the spinal cord that tell muscles in
the arms and legs to respond.

              “By time the brain receives the information, it’s

already out of date,” said Maus.

              The researchers said it can take one-tenth of a second

for the brain to process what the eye sees. That
means, for example, that by the time the brain
“catches up” with incoming visual information, a fast
moving tennis or baseball would already have moved
10 to 15 feet closer than the image in the eye.

              A region in the back of the brain, called area V5,

computes information about motion and position — and
projects where it thinks the ball should be, rather
than where the eyes saw it.

              For the experiment, six volunteers had their brains

scanned with a functional MRI as they viewed the
“flash-drag effect,” a two-part visual illusion in
which we see brief flashes shifting in the direction
of a motion.

              The researchers found that the illusion — flashes

perceived in their predicted locations against a
moving background and flashes actually shown in their
predicted location against a still background —
created the same neural activity patterns in the V5
region of the brain.

              In an earlier study, they stimulated this part of the

brain to interfere with neural activity, and disrupted
this visual position-shifting mechanism.

              The finding could also help explain why altered

trajectories can fool us — such as tennis backspins or
baseball pitches with so-called late break.

              A clearer understanding of how the brain processes

objects in motion can eventually help in diagnosing
and treating myriad disorders, including those that
impair motion perception, according to the UC Berkeley
team. People who cannot perceive motion cannot predict
locations of objects and therefore cannot perform
tasks as simple as pouring a cup of coffee or crossing
a road, researchers said.

              “The brain doesn’t work in real time,” said Maus.                  

              Science

thrives on unanswered questions.

              Religion thrives on unquestioned

answers.

  --

  Richard S. Marken PhD

  rsmarken@gmail.com

  [www.mindreadings.com](http://www.mindreadings.com)

[From Rick Marken (2013.05.27.1245)]

Martin Taylor (2013.05.27.14.49) –

Rick Marken (2013.05.27.1015)–

              [From

today’s Albuquerque Journal, p. C4. I have the Neuron
article if anyone’s interested – Ted]

              FAST

FORWARD

Scientists pinpoint how brain tracks speeding objects

** **** By
Lisa M. Krieger**

** ** ** McClatchy
Newspapers**

              SAN JOSE, Calif. — The human brain is far slower than

a Major League fastball or a blistering tennis serve —
but it has figured out a workaround.

              New research by University of California-Berkeley

scientists solves a puzzle that has long mystified
anyone who has watched, in awe, as elite athletes
respond to incoming balls that can surpass 90 mph.

              The brain perceives speeding objects as further along

in their trajectory than seen by the eyes, giving us
time to respond, according to research by Gerrit Maus,
lead author of a paper published in a recent issue of
the journal Neuron.

      RM: Thanks, Ted. I can't think of a more fitting example of why

PCT is not better known and understood. The “feedback too
slow” myth lives on.

MT: Where do get this from the cited text?

RM: From things like “… elite athletes
respond to incoming balls that can surpass 90 mph.” and " The brain perceives speeding objects as further along
in their trajectory than seen by the eyes, giving us
time to respond". The research described here is clearly based on the idea that hitting a fast ball or a tennis serve is an S-R process; the S is often considered the “sensory feedback” that causes the response. The “mystery” that is “solved” by this research is how information about a speeding ball’s position gets to the response mechanism in time for the appropriate response to be caused since neural conduction rates are slow relative to the speed of the object. This is the “feedback to slow” myth. It makes “responding” to a fast moving object mysterious only if you think people respond to speeding objects. But control theory shows us that we don’t “respond” to speeding objects; we control perceptions that are simultaneously a function of both disturbances (like the speeding object) and outputs (like the swing of the bat or racquet). We control, we don’t react.

Best

Rick

···
    On Mon, May 27, 2013 at 10:02 AM, Ted > > Cloak <tcloak@unm.edu> > >         wrote:
That they say it is

mysterious how it is possible for skilled athletes to perform as
they do? All I can see in the text is that they demonstrated that
the visual system can and sometimes does report a predicted location
of a moving object as though it was the actual position. Whether the
experiment suffices to demonstrate this is something I cannot judge
from the new report, but neither can I see in the news report
anything that suggests “feedback is too slow”.

Methinks you are perceiving what you predict to be in the text.



Martin
              This clever adjustment — compensating for the sluggish

route from the eyes to neural decision-making — “is a
sophisticated prediction mechanism,” he said.

              “As soon as the brain knows something is moving, it

pushes the position of the object moving forward, so
there’s a more accurate measure of where this object
actually is,” said Maus.

              This is useful in survival situations far more

important than sports — such as when we’re crossing a
street, in front of a speeding car.

              Former Yankees catcher Yogi Berra pondered the

mystery, once asking: “How can you think and hit at
the same time?”

              You can’t, because there’s not time for both.
              “But you don’t need to think about it, because the

brain does it automatically,” said Maus.

              At the average major   league speed of 90 mph, a

baseball leaves the pitcher’s hand and travels about
56 feet to home plate in only 0.4 seconds, or 400
milliseconds.

              Tennis is even faster. Last May, courtside radar guns

measured a serve by British player Samuel Groth at 163
mph.

              In that split second, there’s a lot of work for the

body to do. Eyes must first find the ball. The sensory
cells in the retina determine its speed and rush this
information to the brain. Then the brain sends
messages through the spinal cord that tell muscles in
the arms and legs to respond.

              “By time the brain receives the information, it’s

already out of date,” said Maus.

              The researchers said it can take one-tenth of a second

for the brain to process what the eye sees. That
means, for example, that by the time the brain
“catches up” with incoming visual information, a fast
moving tennis or baseball would already have moved
10 to 15 feet closer than the image in the eye.

              A region in the back of the brain, called area V5,

computes information about motion and position — and
projects where it thinks the ball should be, rather
than where the eyes saw it.

              For the experiment, six volunteers had their brains

scanned with a functional MRI as they viewed the
“flash-drag effect,” a two-part visual illusion in
which we see brief flashes shifting in the direction
of a motion.

              The researchers found that the illusion — flashes

perceived in their predicted locations against a
moving background and flashes actually shown in their
predicted location against a still background —
created the same neural activity patterns in the V5
region of the brain.

              In an earlier study, they stimulated this part of the

brain to interfere with neural activity, and disrupted
this visual position-shifting mechanism.

              The finding could also help explain why altered

trajectories can fool us — such as tennis backspins or
baseball pitches with so-called late break.

              A clearer understanding of how the brain processes

objects in motion can eventually help in diagnosing
and treating myriad disorders, including those that
impair motion perception, according to the UC Berkeley
team. People who cannot perceive motion cannot predict
locations of objects and therefore cannot perform
tasks as simple as pouring a cup of coffee or crossing
a road, researchers said.

              “The brain doesn’t work in real time,” said Maus.                  

              Science

thrives on unanswered questions.

              Religion thrives on unquestioned

answers.

  --

  Richard S. Marken PhD

  rsmarken@gmail.com

  [www.mindreadings.com](http://www.mindreadings.com)


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

[Martin Taylor 2013.05.28.12.22]

[From Rick Marken (2013.05.27.1245)]

        Martin Taylor

(2013.05.27.14.49) –

Rick Marken (2013.05.27.1015)–

                        [From today’s

Albuquerque Journal, p. C4. I have the
Neuron article if anyone’s interested – Ted]

                        FAST

FORWARD

Scientists pinpoint how brain tracks
speeding objects


                          By

Lisa M. Krieger**


                          McClatchy

Newspapers**

                        SAN JOSE, Calif. — The human brain is far

slower than a Major League fastball or a
blistering tennis serve — but it has figured
out a workaround.

                        New research by University of

California-Berkeley scientists solves a
puzzle that has long mystified anyone who
has watched, in awe, as elite athletes
respond to incoming balls that can surpass
90 mph.

                        The brain perceives speeding objects as

further along in their trajectory than seen
by the eyes, giving us time to respond,
according to research by Gerrit Maus, lead
author of a paper published in a recent
issue of the journal Neuron.

                RM: Thanks, Ted. I can't think of a more fitting

example of why PCT is not better known and
understood. The “feedback too slow” myth lives on.

MT: Where do get this from the cited text?

                      RM: From

things like “… elite
athletes respond to incoming balls that can surpass 90
mph.” and " The
brain perceives speeding objects as further along in
their trajectory than seen by the eyes, giving us time
to respond".

I don't see how the fact that their studies showed the latter to be

true (and we know the former to be true) can be construed to argue
for your claim that:

          The research

described here is clearly based on the idea that hitting a
fast ball or a tennis serve is an S-R process;

So far as I can see, there is no evidence in the news report either

for or against your statement. Nor can I see any reference there to
a claim that feedback is too slow. All there is is a report of a
neuroimaging study that appears to suggest that we may perceive a
moving object to be where it will be rather than where it is.

Neither PCT nor S-R can get around the fact that there is likely to

be some transport lag between sensory input and the output
mechanism, which is why predictive perception could be useful in
either theory. You yourself long ago showed improved control if
prediction was included (you were surprised that I calculated
correctly that in your case the prediction was for an advance of, if
I remember correctly, about 180 msec, when you had expected it to be
over a second), and I have done several analyses of real and
simulated data that show the same.

If there is a measurable transport lag, inclusion of simple linear

prediction greatly improves the limit of possible control. Not only
that, but actual control sometimes exceeds the precision that is
possible in the absence of prediction. When prediction is based on
both velocity and acceleration, the limit of possible control is
improved still more. That’s the reason my original essays (early
2010 or 2011, I think) on the limits on control were faulty – I had
not included the effects of prediction.

Martin
···
              On Mon, May 27, 2013 at 10:02 > > > AM, Ted Cloak <tcloak@unm.edu> > > >                   wrote:

[From Rick Marken (2013.05.28.1730)]

Martin Taylor (2013.05.28.12.22)–

MT: Where do get this from the cited text?

                      RM: From

things like “… elite
athletes respond to incoming balls that can surpass 90
mph.” and " The
brain perceives speeding objects as further along in
their trajectory than seen by the eyes, giving us time
to respond".

MT: I don't see how the fact that their studies showed the latter to be

true (and we know the former to be true) can be construed to argue
for your claim that:

RM: I don’t believe these studies showed anything like the latter to be true; and I don’t include myself in the “we” who know the former to be true.

          RM: The research

described here is clearly based on the idea that hitting a
fast ball or a tennis serve is an S-R process;

MT: So far as I can see, there is no evidence in the news report either

for or against your statement.

RM: The only evidence I have of their S-R orientation is the statements I quote above. They are assuming that the perception of movement is the stimulus for the appropriate response (swinging the bat or tennis racquet). So they assumed that speed of movement would be a problem and that prediction was necessary. And they interpreted their experimental result as confirming that prediction occurs. But I read the description of their experiment and I don’t see the results as evidence of neural prediction. In their experiment they presented subjects with either actual or apparent ("flash and drag) motion and measured the same level of neural activity in the V5 region using MRI. My PCT oriented interpretation is that the V5 region activity is the output of a motion detection receptive field that puts out the same level of perceptual signal when the input to the field is real movement or non-movement against a moving background.

MT: Nor can I see any reference there to

a claim that feedback is too slow. All there is is a report of a
neuroimaging study that appears to suggest that we may perceive a
moving object to be where it will be rather than where it is.

RM: The “feedback too slow” phrase is just my recasting, into closed loop terms, of their concern about the brain being too slow to respond correctly to a fast moving object.

MT: Neither PCT nor S-R can get around the fact that there is likely to

be some transport lag between sensory input and the output
mechanism, which is why predictive perception could be useful in
either theory.

RM: I have done a lot of modeling of object interception behavior, where the objects are moving pretty quickly, and I get extraordinarily good fits of the PCT model to data without any prediction involved, even when transport lags were included in the model.

MT: You yourself long ago showed improved control if

prediction was included (you were surprised that I calculated
correctly that in your case the prediction was for an advance of, if
I remember correctly, about 180 msec, when you had expected it to be
over a second), and I have done several analyses of real and
simulated data that show the same.

RM: It would be nice if you could find the reference to that. I vaguely remember it. I think we were looking at the difference between tracking targets that were moving randomly or regularly (sinusoidally). I’m pretty sure I didn’t include real prediction in the model; that would involve a lot of calculation. What I probably did was use a future value of the target as the present target value and found that the subject tacking the regular target acted as though the the future value of the target was the input. An actual PCT model of this behavior would have a higher level system controlling for perceiving sinusoidal movement of cursor and target by varying the reference for the cursor position.

MT: If there is a measurable transport lag, inclusion of simple linear

prediction greatly improves the limit of possible control. Not only
that, but actual control sometimes exceeds the precision that is
possible in the absence of prediction. When prediction is based on
both velocity and acceleration, the limit of possible control is
improved still more. That’s the reason my original essays (early
2010 or 2011, I think) on the limits on control were faulty – I had
not included the effects of prediction.

RM: I agree that prediction can improve control but I question whether such prediction is used by living control systems – the low level control systems that are used in hitting a fast moving ball, for instance. I don’t think it is, for a couple reasons. First, I have never had to incorporate prediction into models of object interception; my models act just like people intercepting objects that are moving in “predictable” trajectories (like baseballs) and unpredictable trajectories (like Frisbees). Second, if the brain does do prediction, then that would take some “processing time” (not a factor in nanosecond computer implementations of predictive control systems), which would just add to the transport lag, presumably then requiring prediction father into the future, where predictions would seem to be less useful.

But this article on evidence for predictive control is getting a lot of press. I think that’s because everyone “knows” that hitting fastballs must involve prediction. Just like people “knew” that the sun goes around the earth. All you have to do is look – without putting on your Galilean or PCT glasses.

Best regards

Rick

···


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

[Martin Taylor 2013.05.28.22.09]

In other words you don't believe that elite athletes can hit balls

moving at 90 mph? Is the problem with the radar guns or the batting
statistics? I did a search of my spotty archives, and, lo and behold, I
discovered <Martin Taylor 940204 18:30> in response to
<Rick Marken (940202.1330)>.
--------quote-------

The model that generated stability factors closest to the one’s
observed sampled the equivalent of [ ] ahead. [ Martin: If you
are
reading this, can you intuit, based on IT, how far ahead the
model had to
predict the since disturbance in order to match the subject’s
performance?].
Very roughly, 180 msec according to my calculations. My intuition
would have said about double that. I don’t know which to trust. But
I’ll have to go with 180 msec unless I can find a mistake in the
calculation. You obviously know the right answer, so we’ll see.
--------end quote-------
And then I made a stupid mistake and “corrected” my calculation!
Isn’t that they way of the world?
You are quite right. You were doing a combined compensatory and
pursuit tracking task with two disturbances, one added to the cursor
and one for the disturbed cursor to pursue. You compared sinusoid
and random disturbances. You didn’t use the derivative as I had
remembered (and as I suggest below). You used the actual future
value of the pursued disturbance, but you used it as shown in the
diagram below, as if you were using the derivative – which isn’t
very different when the time-shift is small compared to one cycle of
the disturbance waveform.
No, hardly any calculation at all. My memory (now shown to be
faulty) is that you used the current velocity, the derivative of the
perceptual signal, like this:
That’s mathematically equivalent to what the diagrammed circuit
does, as you said in the messages I found. It’s just easier to show
it as the circuit in the diagram, and that’s the circuit you said
you used in your computation.
If prediction is useful in improving control, it would be strange if
living control systems didn’t use it, don’t you think? The way I became convinced that prediction was necessary in at least
some cases was that Bill’s tracks were either very close to, or
better than, my theoretical best possible control without prediction
for a given lag and disturbance bandwidth, and the best fitting
model was distinctly better than the theory said should have been
possible. By including linear prediction (first derivative only),
the theoretical best possible performance was much better than was
achieved by either Bill or the best fitting model. I don’t agree with the second point. The diagrammed circuit would
not add even a nanosecond to the transport lag. As for the first point, when you say you have not had to incorporate
prediction into your models, I have two comments (1) You have
studied relatively slow interceptions, slow enough that you model
the catcher’s running trajectory where 180msec doesn’t make an
appreciable difference, rather than the fast interceptions involved
in hitting a baseball, or in the kind of catches I used to have to
make when fielding close at cricket, where 180 msec is a good chunk
of the total time available for control. (2) Control without
prediction may be a good match to the human, but unless you test it,
you cannot say whether it might not be a better match with
prediction.
[Aside] Incidentally, on the (unfortunately few) occasions when my
Processing program seems to be working, my otimizations always seem
to show that inclusion of prediction in the manner of the diagram
above does improve the fit to human (my) tracking. As you know from
our private correspondence, I don’t yet quite trust these results,
but, such as they are, they are the results that I have been
getting.
Is it an article on evidence for predictive control? I thought it
was an article about evidence that the perceptual system behaves
neurologically as though we can perceive a predicted position for a
moving object.
I mentioned long ago a personal experience in which I consciously
used predictive perception to win at table tennis played by the
light of the full moon. Initially I would miss the ball because of
playing after the ball had passed me, but I realized quite quickly
that under such low light conditions, perception is somewhat delayed
(see the Pulfrich phenomenon for a demonstration of this effect), so
I began trying to hit the ball at the moment that it appeared to
have reached the net rather than when it appeared to have reached
me. I was very successful against opponents who had not realized the
solution to the delayed perception was to use prediction. We CAN use prediction. Whether we normally do use prediction is a
matter for experiment, but as I said above, wouldn’t it be rather
strange if prediction normally would improve control but is never
used?
Martin

MarkenVelocity.jpg

···

[From Rick Marken (2013.05.28.1730)]

        Martin Taylor

(2013.05.28.12.22)–

                MT: Where do

get this from the cited text?

                                      RM:

From things like “…
elite
athletes respond to incoming balls that can
surpass 90 mph.” and "
The
brain perceives speeding objects as further
along in their trajectory than seen by the eyes,
giving us time to respond".

        MT: I don't see how the fact that their studies showed the

latter to be true (and we know the former to be true) can be
construed to argue for your claim that:

       RM: I don't believe these studies showed anything like the

latter to be true; and I don’t include myself in the “we” who
know the former to be true.

        MT: You yourself long

ago showed improved control if prediction was included (you
were surprised that I calculated correctly that in your case
the prediction was for an advance of, if I remember
correctly, about 180 msec, when you had expected it to be
over a second), and I have done several analyses of real and
simulated data that show the same.

      RM: It would be nice if you could find the reference to that.

I vaguely remember it.

      I think we were looking at the difference between tracking

targets that were moving randomly or regularly (sinusoidally).
I’m pretty sure I didn’t include real prediction in the model;
that would involve a lot of calculation.

      An actual PCT model of this behavior would have a higher

level system controlling for perceiving sinusoidal movement of
cursor and target by varying the reference for the cursor
position.

        MT: If there is a

measurable transport lag, inclusion of simple linear
prediction greatly improves the limit of possible control.
Not only that, but actual control sometimes exceeds the
precision that is possible in the absence of prediction.
When prediction is based on both velocity and acceleration,
the limit of possible control is improved still more. That
turned out to be the reason my original essays (early 2010
or early 2011, I think) on the limits on control were faulty
– I had not included the effects of prediction.

  RM: I agree that prediction can improve control but I question

whether such prediction is used by living control systems – the
low level control systems that are used in hitting a fast moving
ball, for instance.

  I don't think it is, for a couple reasons. First, I

have never had to incorporate prediction into models of object
interception; my models act just like people intercepting objects
that are moving in “predictable” trajectories (like baseballs) and
unpredictable trajectories (like Frisbees). Second, if the brain
does do prediction, then that would take some “processing time”
(not a factor in nanosecond computer implementations of predictive
control systems), which would just add to the transport lag,
presumably then requiring prediction father into the future, where
predictions would seem to be less useful.

  But this article on evidence for predictive control is getting a

lot of press.

[From Rick Marken (2013.05.29.0930)]

Martin Taylor (2013.05.28.22.09)

                MT: Where do

get this from the cited text?

                                      RM:

From things like “…
elite
athletes respond to incoming balls that can
surpass 90 mph.” and "
The
brain perceives speeding objects as further
along in their trajectory than seen by the eyes,
giving us time to respond".

        MT: I don't see how the fact that their studies showed the

latter to be true (and we know the former to be true) can be
construed to argue for your claim that:

       RM: I don't believe these studies showed anything like the

latter to be true; and I don’t include myself in the “we” who
know the former to be true.

MT: In other words you don't believe that elite athletes can hit balls

moving at 90 mph? Is the problem with the radar guns or the batting
statistics?

RM: I
know that elite athletes can hit balls moving at 90mph (though only at a
30% rate, .300 being about the average major league batting average, I believe). What I said that I don’t believe is that these athletes do this by
“responding to incoming balls”. I don’t believe they are responding; I believe they are controlling (though not all that well). One of Bill Powers’ most important contributions to our understanding of the behavior of living systems, I think, was showing us that the appearance that behavior is a response to some stimulus is an illusion as compelling as the illusion that the sun is moving around the earth. It looks like a batter is responding to the balls’ movement but, in fact, the batter is controlling a perception, and it’s the job of the control theory scientist who is trying to understand behaviors like 'hitting a baseball" to figure out what perception that is.

  RM:But this article on evidence for predictive control is getting a

lot of press.

MT: Is it an article on evidence for predictive control? I thought it

was an article about evidence that the perceptual system behaves
neurologically as though we can perceive a predicted position for a
moving object.

RM:
I think they found the evidence of neurological prediction because they
assumed that tasks like hitting a fast ball require prediction. As I said, the description of the results of the study seem much more compatible with the idea that the MRI was showing the output of a motion
detector receptive field than neurological “prediction”.

MT: I mentioned long ago a personal experience in which I consciously

used predictive perception to win at table tennis played by the
light of the full moon. Initially I would miss the ball because of
playing after the ball had passed me, but I realized quite quickly
that under such low light conditions, perception is somewhat delayed
(see the Pulfrich phenomenon for a demonstration of this effect), so
I began trying to hit the ball at the moment that it appeared to
have reached the net rather than when it appeared to have reached
me. I was very successful against opponents who had not realized the
solution to the delayed perception was to use prediction.

RM:
And it sounds to me like what you were doing was controllingthe same
perception (relationship between perceived movement of ball and paddle)
at a different reference level; rather than controlling for the paddle meeting the ball you controlled for the paddle being ahead of the ball.

MT: We CAN use prediction.

RM:
Yes, we can. At cognitive levels. It’s called planning. Which is good but, since we know where the best laid plans aft gang (hint: it starts with “a”), it’s good that we are controlling (not predicting) when we try to carry out those plans.

Best

Rick

···


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

[Martin Taylor 2013.05.29.12.49]

[From Rick Marken (2013.05.29.0930)]

Martin Taylor (2013.05.28.22.09)

                          MT: Where do get

this from the cited text?

                            RM:

From things like "…

                              elite

athletes respond to incoming balls
that can surpass 90 mph." and "

                              The

brain perceives speeding objects as
further along in their trajectory than
seen by the eyes, giving us time to
respond".

                  MT: I don't see how the fact that their studies

showed the latter to be true (and we know the
former to be true) can be construed to argue for
your claim that:

                 RM: I don't believe these studies showed anything

like the latter to be true; and I don’t include
myself in the “we” who know the former to be true.

        MT: In other words you don't believe that elite athletes can

hit balls moving at 90 mph? Is the problem with the radar
guns or the batting statistics?

      RM:                I know that

elite athletes can hit balls moving at 90mph (though only
at a 30% rate, .300 being about the average major league
batting average, I believe). What I said that I don’t
believe is that these athletes do this by
“responding to incoming balls”.

Ah, if I had realized that you were simply responding to the

stimulus “response”, rather than controlling for understanding what
is and what is not intended by the writer of the text, I wouldn’t
have bothered. It seems we are in violent agreement about much (but
not all) of the science of what happens, though not about what can
legitimately be understood from the text.

              RM:But this article on evidence

for predictive control is getting a lot of press.

          MT: Is it an article on evidence for predictive control? I

thought it was an article about evidence that the
perceptual system behaves neurologically as though we can
perceive a predicted position for a moving object.

        RM: I think they found the evidence of

neurological prediction because they assumed that tasks like
hitting a fast ball require prediction.

Maybe they did and maybe they didn't. Is that relevant to the

observations reported?

        As I said, the description of the

results of the study seem much more compatible with the idea
that the MRI was showing the output of a motion detector
receptive field than neurological “prediction”.

How else would "prediction" work? Unless you can detect and quantify

“motion” you can’t do prediction. But I can’t see evidence in the
news report of ONLY the motion detector system being observed, and I
don’t have the original article to check what was actually done and
observed. All I can go on is the report that the observation is what
would be expected if the perceptual system reported the predicted
position.

          MT: I mentioned long ago a personal

experience in which I consciously used predictive
perception to win at table tennis played by the light of
the full moon. Initially I would miss the ball because of
playing after the ball had passed me, but I realized quite
quickly that under such low light conditions, perception
is somewhat delayed (see the Pulfrich phenomenon for a
demonstration of this effect), so I began trying to hit
the ball at the moment that it appeared to have reached
the net rather than when it appeared to have reached me. I
was very successful against opponents who had not realized
the solution to the delayed perception was to use
prediction.

        RM: And it sounds to me like what you

were doing was controllingthe same perception (relationship
between perceived movement of ball and paddle) at a
different reference level; rather than controlling for the
paddle meeting the ball you controlled for the paddle being
ahead of the ball.

Yes.

MT: We CAN use prediction.

  RM: Yes, we can. At cognitive levels. It's called planning. Which

is good but, since we know where the best laid plans aft gang
(hint: it starts with “a”), it’s good that we are controlling (not
predicting) when we try to carry out those plans.

Of course.

I notice that you didn't quote the bits of my message dealing with

matching human track with control models that do and that do not
include prediction.

There are two issues here: "what can you infer from the news report

about what the researchers believe?", and “is prediction
never/sometimes/normally used in control?”. If this debate is to
continue, I suggest the two issues should be separated. I don’t care
about the first, other than insofar as your “knee-jerk S-R response”
to the news report points up one reason it is so hard to get PCT
understood by the mainstream. I do care about the second, because it
is a part of the science of perceptual control.

Martin

[From Rick Marken (2013.05.29.1120)]

Martin Taylor (2013.05.29.12.49)–

MT: I notice that you didn't quote the bits of my message dealing with

matching human track with control models that do and that do not
include prediction.

RM: I had written a reply to that but then managed to somehow unrecoverably delete that text from my message and I was too lazy to try to re-write it. I’m still lazy but I will try to give you a quick summary of what I recall. First, the pursuit tracking task where the cursor was also disturbed makes it impossible to use prediction as a basis of responding since the mouse movements must also compensate for the disturbance to the cursor. So prediction would only help (when there are regular target movements) with the setting of the reference for cursor position. Second, my finding that using the actual future value of the target to set the reference for the cursor does not necessarily mean that that is the result of prediction. It could be control of a higher level perception, such as the relative speed of motion of target and cursor. So I don’t think the results of the tracking studies we discussed back then are evidence for (or against) predictive control. I think what we need is a model of tracking using predictive control and then we can compare it to a model that doesn’t use it and see what kinds of manipulations would allow discrimination between the two.

MT: There are two issues here: "what can you infer from the news report

about what the researchers believe?", and “is prediction
never/sometimes/normally used in control?”. If this debate is to
continue, I suggest the two issues should be separated. I don’t care
about the first, other than insofar as your “knee-jerk S-R response”
to the news report points up one reason it is so hard to get PCT
understood by the mainstream.

RM: Really? This is very interesting to me. Could you explain why you think my "knee-jerk S-R response"

to the news report points up one reason it is so hard to get PCT
understood by the mainstream? I would have thought that it is the mainstream’s S-R conception of how behavior works that makes it so hard to get PCT understood by them.

MT: I do care about the second, because it

is a part of the science of perceptual control.

RM: I think this second issue -- "is prediction

never/sometimes/normally used in control?" – is only tangentially related to the science of PCT. There is no “prediction” in the hierarchical PCT model; apparent prediction, like that seen when a fielder appears to be predicting where a ball will land when he runs to catch it, is just a way of looking at behavior that is organized around the predictionless control of present time perception (note the apparent prediction in my baseball catch demo: http://www.mindreadings.com/ControlDemo/CatchXY.html).

Of course, if prediction is required to understand some behavior then it will have to be incorporated into the PCT model. So in that sense prediction is relevant to the science of perceptual control; it may be involved in control. But prediction is not part of the hierarchical PCT model as it was developed by Powers and I know of no research that requires prediction as a necessary part of the explanation of behavior.

When you can show me the research that demonstrates the necessity of including prediction in a model in order to account for an example of control behavior, then I’ll be interested in looking at prediction as part of the science of perceptual control. But until then I will focus on doing research aimed at determining the most important aspect of behavior from a PCT perspective – the aspect that is ignored by all those psychologists who evoke that knee jerk S-R reaction in me: controlled variables.

Best

Rick

···


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

[From Ted Cloak (2013.05.29.11:58 MST)]

[From Rick Marken (2013.05.29.0930)]

Martin Taylor (2013.05.28.22.09)

MT: Where do get this from the cited text?

RM: From things like “…elite athletes respond to incoming balls that can surpass 90 mph.” and “The brain perceives speeding objects as further along in their trajectory than seen by the eyes, giving us time to respond”.

MT: I don’t see how the fact that their studies showed the latter to be true (and we know the former to be true) can be construed to argue for your claim that:

RM: I don’t believe these studies showed anything like the latter to be true; and I don’t include myself in the “we” who know the former to be true.

MT: In other words you don’t believe that elite athletes can hit balls moving at 90 mph? Is the problem with the radar guns or the batting statistics?

RM: I know that elite athletes can hit balls moving at 90mph (though only at a 30% rate, .300 being about the average major league batting average, I believe). What I said that I don’t believe is that these athletes do this by “responding to incoming balls”. I don’t believe they are responding; I believe they are controlling (though not all that well). One of Bill Powers’ most important contributions to our understanding of the behavior of living systems, I think, was showing us that the appearance that behavior is a response to some stimulus is an illusion as compelling as the illusion that the sun is moving around the earth. It looks like a batter is responding to the balls’ movement but, in fact, the batter is controlling a perception, and it’s the job of the control theory scientist who is trying to understand behaviors like 'hitting a baseball" to figure out what perception that is.

RM:But this article on evidence for predictive control is getting a lot of press.

MT: Is it an article on evidence for predictive control? I thought it was an article about evidence that the perceptual system behaves neurologically as though we can perceive a predicted position for a moving object.

RM: I think they found the evidence of neurological prediction because they assumed that tasks like hitting a fast ball require prediction. As I said, the description of the results of the study seem much more compatible with the idea that the MRI was showing the output of a motion detector receptive field than neurological “prediction”.

MT: I mentioned long ago a personal experience in which I consciously used predictive perception to win at table tennis played by the light of the full moon. Initially I would miss the ball because of playing after the ball had passed me, but I realized quite quickly that under such low light conditions, perception is somewhat delayed (see the Pulfrich phenomenon for a demonstration of this effect), so I began trying to hit the ball at the moment that it appeared to have reached the net rather than when it appeared to have reached me. I was very successful against opponents who had not realized the solution to the delayed perception was to use prediction.

RM: And it sounds to me like what you were doing was controllingthe same perception (relationship between perceived movement of ball and paddle) at a different reference level; rather than controlling for the paddle meeting the ball you controlled for the paddle being ahead of the ball.

MT: We CAN use prediction.

RM: Yes, we can. At cognitive levels. It’s called planning. Which is good but, since we know where the best laid plans aft gang (hint: it starts with “a”), it’s good that we are controlling (not predicting) when we try to carry out those plans.

From Ted Cloak (2013.05.29.11:58 MST)

TC: Here’s my take: The batter is controlling a perception of the bat striking the ball (among a lot of other related and unrelated things). (His perception of) the ball’s movement is an external disturbance of his effort. The article attempts to explain the relation between the true movement of the ball and his perception of same. Over thousands of generations of evolution as a rock-throwing and –dodging species and hundreds of hours of practice, the batter’s perceptual apparatus has been honed to make his ability to control the his desired perception, in the face of that disturbance, fairly efficient (around 30% efficient). That’s why he “sees” the ball not where it is, but where it is going to be.

Where the newspaper article really blows it is in the following: “Then the brain sends messages through the spinal cord that tell muscles in the arms and legs to respond.” (I haven’t read the original Neuron article to see how badly it handles that part, but I don’t imagine it’s any better.)

[From Rick Marken (2013.05.30.1045)]

Ted Cloak (2013.05.29.11:58 MST)–

TC: Here’s my take: The batter is controlling a perception of the bat striking the ball (among a lot of other related and unrelated things). (His perception of) the ball’s movement is an external disturbance of his effort.

RM: The disturbance to the controlled perception is the actual physical trajectory of the ball. The controlled perception is probably something like the relationship between the optical trajectories of ball and bat.

TC: The article attempts to explain the relation between the true movement of the ball and his perception of same. Over thousands of generations of evolution as a rock-throwing and –dodging species and hundreds of hours of practice, the batter’s perceptual apparatus has been honed to make his ability to control the his desired perception, in the face of that disturbance, fairly efficient (around 30% efficient). That’s why he “sees” the ball not where it is, but where it is going to be.

RM: But this idea that we can hit fast balls because we can see the ball where it will be rather than where it is is just an assumption; it is not based on tests of models of hitting balls with bats. Based on my own work on object interception, I am not convinced that hitting a fast ball at the rates people can do it (30%) cannot be accounted for by a plain vanilla control model (which would include a transport lag but no perceptual prediction). It would be a nice exercise for someone to develop a model of hitting a moving object to see how transport lag might affect the level of control as a function of speed of the ball.

As far as the perceptual research described in the article, I still think that interpreting that result as showing perception of a predicted future state of an object is going way beyond the data, to say nothing of ignoring all the great work in neurophysiology that was done by Hubel and Weisel well before these MRI jockey’s came on the scene. Hubel and Weisel found that individual optic neurons respond selectively to different kinds of sensory stimulation; some only fire when there is an edge of a particular orientation; some only respond to movement in a particular direction,etc.In other words, individual neurons seem to be connected to perceptual functions (called receptive fields) that are “looking for” particular features in the optic array. For this work they won the Nobel in physiology. This picture of how the brain perceives the world is very compatible withthe PCT view of neural perceptual processing.

The work described in the subject article seems perfectly consistent with the findings of Hubel and Weisel, except that the MRI (as far as I know) measures firing in a collection of neurons rather than in a single neuron, as in the Hubel/Weisel work. The subject paper says that both real and illusory movement produce the same level of activity in the same brain region. This sounds to me like the output of a movement detector receptive field. Calling it prediction is, as I said, reading stuff into the data that just ain’t there.

TC: Where the newspaper article really blows it is in the following: “Then the brain sends messages through the spinal cord that tell muscles in the arms and legs to respond.” (I haven’t read the original Neuron article to see how badly it handles that part, but I don’t imagine it’s any better.)

RM: I’m sure it’s not any better. This quote – “Then the brain sends messages through the spinal cord that tell muscles in the arms and legs to respond” – probably comes directly from the authors. These newspaper reports of scientific findings are based on press releases from the journal (which are typically approved by the author(s) of the article) and/or interviews with the authors (and/or people familiar with their work). So, as I said to Martin, this dreadful piece of research was based on the idea that batting is an S-R process: ball movement is the stimulus and the brain sends a message about the state of this stimulus to the response system – but sends it too late so it has to send a message about the predicted future state of the stimulus. These folks are just doing what psychologists do now to try to look scientific – use high tech physiological measuring equipment to make it seem like they have discovered something important. It would be depressing if it weren’t so amusing.

Best

Rick

···


Richard S. Marken PhD
rsmarken@gmail.com

www.mindreadings.com