[From Rick Marken (930328.1500)
CHUCK TUCKER (930328)
Some time ago Rick (930315.1500) stated that his research in "Mind Readings"
Chapter 3 ("The Cause of Control Movements in a Tracking Task" proved
"... THERE IS NO INFORMATION ABOUT DISTURBANCE IN THE PERCEPTUAL SIGNAL
CONTROL SYSTEM. This means that THE PERCEPTUAL INPUT TO A CONTROL SYSTEM
CANNOT BE WHAT CAUSES THE OUTPUR OF THE SYSTEMN (sic) TO MIRROR THE
DISTURBANCE." Reading his article again I don't believe that he
proves that al all
Good. This is all I want; look at the research. (By the way, did I use
the word "prove" -- it's not in quotes above? If I did, I'm sure
I meant it to be synonymous with "test" rather than "deductive proof").
So why doesn't my experiment "prove" this to your satisfaction? You say
it's because:
the subjects did have an awareness of
a disturbance - they were told about it and through several test runs
of the task knew that a disturbance was influncing the cursor - it is
the case that they did now know WHAT is was (neither did Marken - it was
random).
Two problems here: 1) they were not told about the disturbance -- though
they could surely tell that it was present and 2) it was a sine wave,
not random (though it works just as well with a filtered random
disturbance).
But I don't see what either of there points (subject's being aware of
the disturbance and the disturbance being random) has to do with the
point of the paper: that there is no information in the cursor movement
that can be (or is being) used by the subject to determine how to make their
responses? I assume "information about the disturbance" means that there
is some something about cursor movement (p(t)) that allows the subject
to compute o(t) -- ie. o(t) = f(p(t)) -- so that o(t) = -kd(t). The
experiment you read showed, I think, that there is no f() that could be
turning p(t) into o(t) because you get virtually the same o(t) twice
when you use the same d(t) but p(t) is totally different on each occassion.
I would have refused to published this study since WHAT the subjects
were told is not to be found in this description BUT having done
this type of work I know that they had to be told something (unless all
of the subjects had done this many times) BUT IT SEEMS OBVIOUS TO ME
THAT THE SUBJECTS WHERE AWARE THAT A DISTURBANCE WAS INFLUENCING THE
CURSOR POSITION ALONG WITH THEIR MOVEMENT OF THE GAME PADDLE. If this
investigator believes otherwise let him present evidence that they
were ignorant of a disturbance. He should has ASKED THEM or used THE
TEST.
Does this study upset you, Chuck?
Why would the subject's being aware or not aware of the disturbace
make any difference? The subject can certainly tell that there is
a disturbance operating and I would have reported that fact if
I (or anyone) could have seen why it would be germain. What is
far more important is "did the subject have any idea about the nature
if the function d(t)". In fact, one clever reviewer (Tom Bourbon, it
was, before he fell into the PCT all the way) pointed out that,
because I used sine waves, the subject could tell from their own
hand movements what the disturbance function was. Tom suggested that
the subjects could be producing the sinusoidal outputs from memory.
This was a far-fetched, but at least pertinent, criticism of my study.
I pointed out that the phase and frequency would have been impossible
to predict on the repetition trials (which were non-continguous and
the subject had no way of knowing when the d(t) on a particular trial
was a repetition of the d(t) on a previous one-- but, still, I should have
done the study with random disturbances. In fact, I have done the study
(just the other day) with random disturbances and it works just the same.
So it looks like your only complaint about the study is that the
subjects were aware of the disturbance; in fact, they would be
aware of the distrubance whether I told them about it or not; they
know that they have to do something to keep the cursor on target.
I think it would be hard to keep the subject "unaware" of the
existence of the disturbance. So I guess I have to ask how this
awareness might affect the results -- ie, the fact that the correlation
between output traces on two (very separate) trials with the same
disturbance was typically .99+ while the correlation between p(t)
(the only thing the subject could see on those trials) was typically
less than .2, once as low as .0032?
Best
Rick