[From Bill Powers (980128.0637 MST)]
Here's the post that some people apparently didn't get:
···
---------------------------------------------------------------
[From Bill Powers (980126.1043 MST)]
Bruce Abbott (970126.0940 EST)--
The really interesting thing about this is that we somehow infer a stable
frame of reference outside ourselves, one within which we move.
Yes. And even more interesting than that is that this stable frame of
reference must actually exist in our brains: "Outside" is also a construct of
the brain. There are some truly delightful puzzles that will be involved in
working all this out.
But try the following. ...
These are all good data for a model of spatial perception. In my model of
oculomotor control, all these illusions are explained as the operation of two
basic control systems: a pursuit tracking system and a positioning system
that works only during saccades. I'm thinking of working this model up for
presentation at the German CSG meeting in June, since there will be a number
of people interested in visual systems there. While there's a lot that's
conjectural about the model, it does seem to explain the various illusions in
a compact way.
There's another one that's interesting. A stalk is mounted on a contact lens
attached to the eye. When a string is used to pull sideways on the stalk
while the subject fixates on an object straight ahead, the field of vision
seems to swing around in space in the other direction, as in your description
of Helmholz's experiment. But in this case the effect is so strong that
subjects get the feeling they are looking out the backs of their heads -- a
very unpleasant experience! Also, if the stalk is fastened so it can't move
and the person tries to look to the left or right, the whole world seems to
swing violently in the opposite direction, even thought the image on the
retina doesn't move.
Some years ago, Wayne Hershberger raised the question of how people define
"straight ahead." This is what led me to try to build a model of the
oculomotor system. While playing with this question, I came up with a series
of demos, which I briefly described a day or two ago. Here's a more extended
description.
Step 1. Stand in the middle of a room looking straight ahead at some object.
Then move your eyes to some object off to one side (by 20 degrees or so),
without moving your head. That direction is "to the right."
Step 2. Return your eyes to the straight ahead object. Now turn your head
approximately toward the object off to the right, while keeping your eyes and
body pointing straight ahead.
Step 3. Return your head to the straight ahead position. Now, without moving
your head, eyes, hips, or feet, turn your shoulders toward the object on the
right.
Step 4. Return your shoulders to straight ahead. Now, holding your eyes,
head, shoulders, and feet in the straight ahead position, turn your hips
toward the object to the right.
Step 5. Return your hips to straight ahead. Now, holding hips, shoulders,
head, and eyes straight ahead, pivot on your heels so your feet point toward
the object on your right. You can keep your balance if you just sort of hitch
the balls of your feet and toes to the right in a quick move.
At this point only your feet are directed to the right.
Step 6. Holding shoulders, head, and eyes straight ahead, turn your hips to
the right.
Step 7. Holding head and eyes straight ahead, turn your shoulders to the
right.
Step 8. Holding the eyes straight ahead, turn your head to the right.
Step 9. Turn your eyes to the right.
Now your whole body is pointed to the right of the original straight-head
position. With a mere mental twitch (which you could have performed at any
time during the previous steps, and which may happen involuntarily) you can
now redefine the direction in which your body is oriented as "straight
ahead," and it will be so. The object that was off to the right is now
straight ahead.
We seem to have here a nice stack of control systems, each physically mounted
on the one below it. And the "objective" frame of reference can change at the
observer's whim. This stack is highly amenable to modeling with control
systems, and the perceptions involved interact in interesting ways (all
starting at the same level -- this is not hierarchical control). It was this
sort of fooling around that gave me the idea for the "oculomotor" model,
which actually extends in a nice regular way to the whole stack. You can
perform a "foot saccade" just as easily as an eye saccade.
Best,
Bill P.