Makoto HIRAYAMA, Mitsuo KAWATO, Michael I. JORDAN
Feedforward Neural Network Modeling
of Target-directed Arm Movement
which Reproduces Speed-Accuracy Trade-off
Abstract:Various properties of the cascade neural network as a computational
model for motor control of a multi-joint arm are studied. The cascade
neural-network model calculates the trajectory based on minimum-torque-change
criterion. If the weighting parameter of the smoothness criterion is
fixed and the number of relaxation iterations is rather small, the cascade
model cannot calculate the exact torque, and the hand does not reach the
desired target using the feedforward control alone. Thus, one observes an
error between the final position and the desired target location. By
simulating target-directed arm movements using a fixed weighting parameter
value and a limited iteration number, we found the cascade model reproduced
the planning time-accuracy trade-off, and speed-accuracy trade-off of the arm
movement, well known as Fitts's law. This work provides a candidate of
possible neural mechanism which explains the stochastic variability of the
time course of the feedforward motor command along with several invariant
features of multi-joint arm trajectories such as roughly straight hand paths
and bell shaped speed profiles.