[From Rick Marken (970505.0950 PDT)]
I have revised the "Hierarchy of Control and Perception" demo
at http://www.leonardo.net/Marken/demos.html
It is now possible to control either one of two possible perceptions
of an environmental variable; you can control either the direction
of apparent motion of letters moving across the screen or the order
(sequence) in which the letters occur. Your ability to control each
perception depends on the speed with which the letters appear on the
screen. Movement can be seen (and controlled) at about 5 letters/sec.
Sequence can be seen (and controlled) at about 2 letters/sec.
One interesting thing about this demo is that it shows the difference
between a transition perception (movement) and a sequence perception
(order in which the letters occur). When you can see one letter
happening after another (movement) you still can see (and control)
the order in which the letters occur.
I think it's interesting that, at the rate at which you can see
letter movement (5 letters/sec) you can see each of the letters
and you can see that they are occuring in different positions but
you can't quite see their order yet. Assuming that it takes more
time to compute higher level than lower level perceptions, it makes
sense (in terms of the PCT hierarchy) that it takes longer (slower
letter speed) to perceive sequence (letter order) than transition
(movement).
Anyway, if anyone actually tries this demo I would appreciate any
comments or suggestions.
Thanks
Rick
[From Bruce gregory (970505.1700 EDT)]
Rick Marken (970505.0950 PDT)
Anyway, if anyone actually tries this demo I would appreciate any
comments or suggestions.
I like the demo, but I can't control sequence worth a damn!
Bruce
[From Rick Marken (970506.0800 PDT)]
Bruce Gregory (970505.1700 EDT)--
I like the demo, but I can't control sequence worth a damn!
Actually, this is quite interesting. David Goldstein was interested
in this demo in terms of its possible implications for dyslexia
(which seems to turn on an inability to perceive sequence). Do you
still have a problem controlling the sequence if you reduce letter speed
to about 1/sec?
Best
Rick
[From Bruce Gregory (970507.1115 EDT)]
Rick Marken (970506.0800 PDT)
Bruce Gregory (970505.1700 EDT)--
> I like the demo, but I can't control sequence worth a damn!
Actually, this is quite interesting. David Goldstein was interested
in this demo in terms of its possible implications for dyslexia
(which seems to turn on an inability to perceive sequence). Do you
still have a problem controlling the sequence if you reduce letter speed
to about 1/sec?
I'm not ignoring you, just trying to improve my virtually
nonexistent tracking skills. Your server seems to be down at the
moment. but I'll try again later.
Bruce
[From Bill Powers (970507.1536 MST)]
Bruce Gregory (970507.1600 EDT)--
I found that a run with Rick's demo lasts about 18 seconds on my computer.
The lowest RMS error I was able to generate was 8.8, with a stability factor
of about 12. This is not seriously different from your 14 RMS. It took me
about 10 runs to achieve this, and my initial RMS errors were in the 20s.
One problem is the slow display rate -- there's a lag that makes it hard to
get good control (I estimate 1/4 to 1/5 second per iteration). Another is
learning how to avoid putting in a large initial transient when the run
starts. Even a few initial bobbles can greatly increase the RMS error. I
think Rick's program has a few seconds' delay before data taking starts, to
allow getting control (if not, it should). It pays, therefore, to get
control as fast as possible. Just move the mouse arrow up enough to get it
off the run button and immediately start tracking.
I'll bet you can get better with practice. Not that it matters -- we don't
give prizes for beating any records.
But as you surmise, mastery may be its own reward (isn't it amazing how many
folk sayings translate directly into PCT -- and how few scientific
descriptions do?).
Best,
Bill P.
[FRom Bill Powers (970507.1523 MST)]
Bruce Gregory (970507.1600 EDT)--
My RMS error for The Nature of Control hovers between 15 and 20.
The best I have been able to do is 14.3, far above the 4.0
target. My future as a compensatory tracker looks bleak...
I suggest that you time the duration of the run with a stopwatch. If the
program is running faster than it should on your machine, the difficulty
will be increased. I'll do the same thing so we can compare notes.
Best,
Bill P.
[From Bruce Gregory (970507.1755 EDT)]
Bill Powers (970507.1523 MST)
I suggest that you time the duration of the run with a stopwatch. If the
program is running faster than it should on your machine, the difficulty
will be increased. I'll do the same thing so we can compare notes.
A run on my machine takes 15 seconds.
Bruce
[From Rick Marken (970507.1515 PDT)]
Bill Powers (970507.1523 MST)
I suggest that you time the duration of the [Nature of Control]
demo run with a stopwatch.
Good idea! It takes 20 seconds on my Power Mac 7600/120. My
RMS error on the run was an apparently clunky 13.4 but my
stability measure was a towering 19.4.
Best
Rick
[From Rick Marken (970507.1540)]
Bill Powers (970507.1536 MST) --
I found that a run with Rick's demo lasts about 18 seconds on my
computer.
The lowest RMS error I was able to generate was 8.8, with a
stability factor of about 12.
The RMS errors I reported previously were wrong. I consistently
get an RMS near 6.5 and a stability near 15.2.
Given the relative speeds of Bill's (18 sec run), Bruce G.'s (15 sec
run) and my machine (20 sec run) it looks like we are all controlling at
about the same level.
Even a few initial bobbles can greatly increase the RMS error. I
think Rick's program has a few seconds' delay before data taking
starts, to allow getting control (if not, it should).
It does. It throws away the first 25 samples.
Love
Rick
[From Bruce Gregory (970507.1600 EDT)]
Rick Marken (970506.0800 PDT)
My RMS error for The Nature of Control hovers between 15 and 20.
The best I have been able to do is 14.3, far above the 4.0
target. My future as a compensatory tracker looks bleak...
Bruce
[From Bruce Gregory (970508.0950 EDT)]
Bill Powers (970507.1536 MST)
I'll bet you can get better with practice. Not that it matters -- we don't
give prizes for beating any records.
Damn. Boy, was I misinformed. I thought the PCT pecking order
was determined by tracking skills. Oh, well...
But as you surmise, mastery may be its own reward (isn't it amazing how many
folk sayings translate directly into PCT -- and how few scientific
descriptions do?).
Funny, I am continually amazed by this as well.
Bruce
[From Rick Marken (970526.1330 PDT)]
I have put a new and (I think) considerably improved version of
the hierarchical control demo (which I now call "A Hierarchy of
Perception and Control") and write-up on my web site. You can get there
directly by going to:
http://www.leonardo.net/Marken/ControlDemo/HP.html
This demo has been a bear to develop because it works onthe basis of the
speed of the animation, and the speed of the animation differs
considerably depending on whether Java is run through the browser or
through the "Applet viewer" (that lets you run Java without going
through the browser). I fear that the speed of animation (and, hence,
the effectiveness of the demo) may also depend stongly on the type of
platform on which it is run (even if it is always run through the
same browser -- say Netscape 3.0 -- on every platform). So I would
appreciate hearing how the demo works for people who are running
it on PCs or Suns (or whatever). I would also appreciate any comments or
suggestions you might have regarding the demo itself and/or the
write-up.
Thanks
Rick
ยทยทยท
--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: marken@leonardo.net
http://www.leonardo.net/Marken
[From Bill Powers (970526.1812 MDT)]
Rick Marken (970526.1330 PDT)
I have put a new and (I think) considerably improved version of
the hierarchical control demo (which I now call "A Hierarchy of
Perception and Control") and write-up on my web site.
Ingenious. Difficult, but learnable. It's particularly interesting to see
the "percent on perception" numbers change as you control the different
levels of perception.
This is most evident if you leave the speed on "slow" and just control
sequence, then transition (direction of movement), then configuration on
successive runs. Evidently you've set it up so that it's not possible to
control a higher-level variable just by controlling a lower-level one in
the same way each time -- that is very clever, and possibly even deep.
This may be the first experiment we have that shows multiple levels of
control in a single experiment. Well done.
Best,
Bill P.