Arms and stuff

[From Bill Powers (930312.0830)]

Avery Andrews (930312.1912) --

Do you have a copy of Arm Version 2.0? It allows you to set up
two target positions, and then have the target jump back and
forth (instantly) between them at about 1-sec intervals, while
the arm tracks. The initial error for each movement is as large
as it can get, the entire magnitude of the target jump. The
trajectory of the fingertip for such movements is a not-quite-
straight line from the initial position to the final position (if
the jump is 5 cm, the deviation from straight is roughly 2 or 3
mm). This is comparable to path curvatures in experiments with
real human arms.

Bizzi et. al. claim to have evidence that hand movements
typically follow pre-determined trajectories ...

In what sort of situation? And what is doing the pre-determining?
In Arm v.2, the paths for maximum-speed jumps are indeed
predetermined, but they're not precalculated -- they simply
emerge from the time-constants of the control systems and the
dynamical properties of the arm. The trajectories for fast
movements are often cited as evidence of a CPG, and their
repeatability is taken as a sign that they're planned. But
neither is necessarily true, or even possible.

For movements taking an appreciable length of time, the control
systems simply follow the reference signals, and any path is
possible. Put a dot on a piece of paper, and another dot about 3
inches to the right of it. Put a pencil on the first dot. Now
write your name cursively so that the end of the last letter
leaves the pencil on the second dot. How's that for a
"predetermined" or "planned" trajectory?



(`deciding what to do is easy when you
know what's in front of you' is the way David Chapman put it).

Oh, cute. How easy is "knowing what's in front of you?"

Bill P.