Walking toward target

[From Bill Powers (990418.1259 MDT)]

Thinking over how a person walks toward a target, I propose this model for
testing:

Model: a person will orient the body and head so the "stack" of
legs-hips-shoulders-head-eyes is aligned in the direction of walking and
the target is seen as foveated, straight ahead.

If glasses are worn that deviate the apparent direction of the target by x
degrees (positive x is a deviation of the target to the right), the person
will still walk with the legs, body, head, and eyes aligned straight ahead
and the apparent target foveated. This will require orienting the body at
an angle -x degrees deviated from the actual direction of the target
relative to the body's straight-ahead vector. Walking toward the apparent
target causes the body to deviate from the actual line to the target by -x
degrees. This deviation is maintained constant when the direction of
walking is adjusted to keep the target apparently straight ahead. The
actual path will therefore be a spiral that converges on the target. The
spiral will be the path for which the radius vector from target to person
is always inclined x degrees to the direction of motion.

This, you say, is what they actually observed. But their conclusion that
direction of walking is guided by retinal location of the target is clearly
false. The direction of walking is continuously changing. The path,
however, is a constant-angle spiral, meaning that the apparent direction to
the target is constant. If the direction of walking were being quided by
the constant retinal position of the target, it should be constant, not
changing.

In fact we see that the continuously changing direction of walking is
exactly what is required to maintain a constant appearance of the target in
the straight-ahead direction. Thus the variation in the walking direction
is controlling the apparent position of the target.

Note that if the deviation angle were 90 degrees, the person would walk in
a circle around the target. This is easy to check with a hand-mirror.

Best,

Bill P.