I did
find one paper that seems to be a precursor of his alcohol studies
(British J. Psychol, 1965, 56, 233-242). It uses the same kind of task
and shows examples of the J-curve,
What's a J curve?
You might like, however, a few words from the summary:
And from the introduction: "A hypothetical function of proprioception is
to provide negative feedback which ensures that a movement initiated by
an error-stimulus reduces that error."
The phrases "provide negative feedback" and "movement initiated by
an error-stimulus" lead me to suspect that this research is based on
the good old input-output model (using feedback terminology) -- the
same input-output model we see in manual control research.
It seems to me that it would be worthwhile to try to gather data akin
to Gibbs, and see how consistent the J-curve effect might be, and
to model it if it is consistent enough.
I always like suggestions for experiments. I would really appreciate
it if you could supply enough detail about the study so that I could
implement a version of it. By the way, I hope the J-curve effect doesn't
depend on drinking a bunch of alcohol; then I couldn't be a subject
in the study (even though I'd REALLY like to).
[Martin Taylor 931122 12:20]
(Rick Marken 931119.1100)
What's a J curve?
Sorry, I forgot that the J-curve discussion was on the net some time ago,
and the recent version was private.
Gibbs did experiments in which the subject had a pointer (sometimes eye
position, sometimes arm on a pivoted rest, as I remember it). The
subject was supposed to point as quickly and accurately as possible at
whichever of 5 lights arranged in a row came on. If the light that had
been on was number 1, 3, or 5, the initial movement was correctly directed
and quickly setlled near the "correct" location. But if the transition
was 2->1 or 4->5, the initial movement was "often" (my word) in the wrong
direction, quickly corrected to move in the right direction. The visual
look of the initial part of the trace of pointer position was J-like.
The correction was far faster than "a visual reaction time." Gibbs
claimed that the initial move was a feedforward action based on the
probability that the next light would be in the majority direction (to
see that a light has come on is quicker than to see where it has come on).
Most of the movement was under negative feedback control, according to
Gibbs. But the very brief initial movement was not. Without having
found the later papers, I assume that he believed that this joint action
occurred whether the initial move was in the right or the wrong direction,
but it could be seen only when the J-curve showed up.
Check out the reference I provided. I think it was the first one showing
J-curves, as I judge from the titles in a bibliography search I did for
his work.
if the transition
was 2->1 or 4->5, the initial movement was "often" (my word) in the wrong
direction, quickly corrected to move in the right direction. The visual
look of the initial part of the trace of pointer position was J-like.
The correction was far faster than "a visual reaction time." Gibbs
claimed that the initial move was a feedforward action based on the
probability that the next light would be in the majority direction (to
see that a light has come on is quicker than to see where it has come on).
Doesn't this sound like control of (anticipated) sequence being corrected
less rapidly than control of (out-of-sequence) position? Parallel the
tracking experiments when the relation of handle movements to cursor
movements is reversed.