Bimanual Control

CSGnet Archive CSG2004
1

[From Rick Marken (2004.07.07.0845)]

My interest in bimanual (two handed) control was rekindled by my discovery
that Franz Mechsner, a biologist at the Max Plank Institute whose work is
comfortably compatible with PCT, has a target article coming out in the
_Journal of Motor Behavior_. Unfortunately, I found out about the article
well after the deadline for contributing commentaries. But it got me to
revisit my efforts to build a model that accounted for the results of one of
the ingenious experiments reported by Franz (in _Nature_) a couple of years
ago. Franz describes that and several other experiments in the JMB article,
which gives an overview of his very impressive research program. Franz's
research is aimed at demonstrating that behavior is organized around the
control of perceptual inputs rather than the calculation of motor outputs.
Franz knows about the relevance of PCT to his work (probably because a
fellow named Scott Jordan, who, I believe was a student of Wayne
Hershberger's and was at one of the early CSG Meetings, has been working in
Franz's lab) but he has not built any working models to demonstrate how well
PCT can explain his results. So I have taken it upon myself to do this, with
some success.

I have tuned up my java model of Franz's bimanual control experiment and put
it up on the web, along with a diagram of the model, at

http://www.mindreadings.com/Coordination.html

This is a work in progress that might turn into a paper. I am putting it up
on the web in order to solicit suggestions, requests for clarifications and
so on. I'd like to see if we can use CSGNet for what I thought was it's
original purpose: to allow PCT researchers, separated geographically, to
discuss their work.

By the way, I am renting a car in Chicago so I will be happy to take some
people to the conference center ( probably will have room for 2 at most) if
they need a ride. I'm coming in at 5:00 PM on Wednesday.

Best regards

Rick

···

--
Richard S. Marken
MindReadings.com
Home: 310 474 0313
Cell: 310 729 1400

2

[Bruce Gregory (2004.070704.1500)]

Rick Marken (2004.07.07.0845)

I have tuned up my java model of Franz's bimanual control experiment
and put
it up on the web, along with a diagram of the model, at

Bimanual Coordination

Very nice! However, the data presentation is on my screen for two short
a time for me to read it. I'm accessing the site with Safari on a Mac.

People believe whatever they need to believe. The longer they believe
it the more they need it.

3

[From Bill Powers (2004.07.07.1252 MDT)]

Rick Marken (2004.07.07.0845)]

My interest in bimanual (two handed) control was rekindled by my discovery
that Franz Mechsner, a biologist at the Max Plank Institute whose work is
comfortably compatible with PCT, has a target article coming out in the
_Journal of Motor Behavior_.

Your model looks good, but perhaps some added work would help. How did you
generate the periodic cranking movements? I suggest using a "central
pattern generator," which would be an oscillator producing sine and cosine
waves. It can be programmed this way:

dx := freq*(y + dy);
dy := -freq(x + dx);

"Freg" determines the frequency of oscillation, so this implements an
oscillator with a controllable frequency. Note that y and x will be 90
degrees out of phase, which is what you need to generate a cranking movement.

To detect the phase difference between two oscillating variables v1
and v2, use the value of the product v1*v2, smoothed. When v1 and v2 are
90 degrees out of phase, the average product will be zero. When they're in
phase, the average product will be positive; out of phase, negative.

Suddenly the meeting is looming. I'll answer Zhang as soon as I can talk
with Mary about our plans for Wednesday.

Best,

Bill P.

4

[From Dick Robertson,2004.07.07.1`500CDT]

···

----- Original Message -----
From: Richard Marken <marken@MINDREADINGS.COM>
Date: Wednesday, July 7, 2004 10:44 am
Subject: Bimanual Control

[From Rick Marken (2004.07.07.0845)]

My interest in bimanual (two handed) control was rekindled by my
discoverythat Franz Mechsner, a biologist at the Max Plank
Institute whose work is
comfortably compatible with PCT, has a target article coming out
in the
_Journal of Motor Behavior_. Unfortunately, I found out about the
articlewell after the deadline for contributing commentaries. But
it got me to
revisit my efforts to build a model that accounted for the results
of one of
the ingenious experiments reported by Franz (in _Nature_) a couple
of years
ago.

Nice demo, Rick and how nice to hear somebody working in parallel without trying to convert PCT to a worn out theoryl

Best,

Dick

5

[From Rick Marken (2004.07.07.1410)]

Bruce Gregory (2004.070704.1500)]

Rick Marken (2004.07.07.0845)

I have tuned up my java model of Franz's bimanual control experiment
and put
it up on the web, along with a diagram of the model, at

Bimanual Coordination

Very nice! However, the data presentation is on my screen for two short
a time for me to read it. I'm accessing the site with Safari on a Mac.

Thanks. Try it in Explorer, if you can. The data display should remain on
the screen until you push the "Symmetry" or "Antiphase" button to run the
model again. I don't know why it doesn't work this way in Safari. I guess I
should download it and see. It runs fine in Explorer on both the Mac and PC.

Was the write-up clear? I put it up rather hastily in preparation for
writing a comment on the Mechsner target article but then I found out, 2
days later, that I was too late. But I want to keep working on this anyway.
The bimanual task developed by Mechsner is, I think, a very elegant
demonstration of control of perception. It is really almost impossible for
people to move their hands simultaneously with different frequencies of
rotation. But they do it just fine (as the model does) when this is the
action required to maintain a perception of symmetrical flag rotation.

Best

Rick

···

--
Richard S. Marken
MindReadings.com
Home: 310 474 0313
Cell: 310 729 1400

6

Phil Runkel to Rick Marken's of 2004.07.07.1410:

You wrote: "It is really almost impossible for
people to move their hands simultaneously with different frequencies of
rotation. But they do it just fine (as the model does) when this is the
action required to maintain a perception of symmetrical flag rotation."

Like riding a bicycle. If you think about how to move the handlebars,
you fall over.

7

[From [From Rick Marken (2004.07.08.0900)]

···

Rick Marken (2004.07.07.0845)]

My interest in bimanual (two handed) control was rekindled by my discovery
that Franz Mechsner, a biologist at the Max Plank Institute whose work is
comfortably compatible with PCT, has a target article coming out in the
_Journal of Motor Behavior_.

Your model looks good, but perhaps some added work would help. How did you
generate the periodic cranking movements? I suggest using a "central
pattern generator," which would be an oscillator producing sine and cosine
waves. It can be programmed this way:

dx := freq*(y + dy);
dy := -freq(x + dx);

"Freg" determines the frequency of oscillation, so this implements an
oscillator with a controllable frequency. Note that y and x will be 90
degrees out of phase, which is what you need to generate a cranking movement.

To detect the phase difference between two oscillating variables v1
and v2, use the value of the product v1*v2, smoothed. When v1 and v2 are
90 degrees out of phase, the average product will be zero. When they're in
phase, the average product will be positive; out of phase, negative.

Suddenly the meeting is looming. I'll answer Zhang as soon as I can talk
with Mary about our plans for Wednesday.

Best,

Bill P.

--
Richard S. Marken
MindReadings.com
Home: 310 474 0313
Cell: 310 729 1400

8

[From Rick Marken (2004.07.08.0900)]

Oops. Sent that previous one by accident. Trying to do too many things at
once.

Bill Powers (2004.07.07.1252 MDT)

Your model looks good, but perhaps some added work would help.

Yes, indeed. First, let me say that I tried running the demo in Safari (the
browser Bruce Gregory used) and it worked like a charm. In fact, all the
demos at my demo site -- Redirect --
including the "Hierarchy of perception and control", which depends on
timing commands that are apparently done correctly using the java built into
Safari, work beautifully in Safari. I suppose Safari only runs on Macs so
that's just tough luck for you PC types;-)

How did you
generate the periodic cranking movements?

By continuously increasing the reference for frequency of movement of the
left hand flag.

I suggest using a "central
pattern generator," which would be an oscillator producing sine and cosine
waves. It can be programmed this way:

dx := freq*(y + dy);
dy := -freq(x + dx);

I like the control of perception of frequency approach better, and it's a
better representation of what the subjects were asked to do, which was to
slowly increase the speed of rotation of the flags. Actually, what was
controlled was da/dt, the size of the change in angle at each instant -- a
transition perception.

To detect the phase difference between two oscillating variables v1
and v2, use the value of the product v1*v2, smoothed.

That's a good idea but I'm using Mechsner's approach to measuring the phase
difference so that I can compare the behavior of the model to that of the
participants in his study. Mechsner measured the angular difference between
the right and left flags on each rotation when the left flag was pointing
due east. So the distributions are counts of the number of times each
angular difference was observed for each rotation. The model seen in the
demo goes through about 800 rotations but only the last 20 or so (the one's
done at the fastest rotation speed) are shown.

By the way, control of the relative phase of flag movement is not perfect in
the demo because I add normally distributed noise to the perception of the
relative phase of flag rotation in proportion to the magnitude of this
perception (representing the size of the phase difference). This is why the
performance (measured in terms of the distributions of phase difference on
each rotation) is better in the symmetry condition (phase difference
typically close to 0) than in the antiphase condition (phase difference
typically close to 180).

Thanks for all the comments so far. I'll keep working on it.

Best regards

Rick

···

--
Richard S. Marken
MindReadings.com
Home: 310 474 0313
Cell: 310 729 1400