We studied the relations between arm posture and hand location as monk
eys performed tasks involving three-dimensional arm movements. Two spe
cific questions were addressed: how reliable are these relations from
trial to trial and from day to day, and are the arm postures dictated
only by the location of the hand or are the postures also dependent on
the task being performed? Based on the variable errors in linear regr
essions, we found that monkeys performed the tasks in a very stereotyp
ic fashion from trial-to-trial: for a given monkey and task, knowing t
he position of the hand was sufficient to determine the posture of the
arm. These relations did not change from day to day: the relation bet
ween hand location and posture was stable over the course of the exper
iments. In some of the tasks, the postures employed by the monkeys wer
e strikingly similar to those observed in human psychophysical studies
. The relations between hand location and arm posture exhibited a larg
e degree of task dependence. Changing the required orientation of the
hand resulted in changes in the arm posture.