Varying the size or stiffness of an external load often results in sys
tematic grasp force or finger span matching errors, typically judged v
ia contralateral matching. In the present study, subjects squeezed com
pliant or rigid manipulanda in each hand using three-finger pinch in o
rder to generate, and simultaneously match, a reference force indicate
d by a visual cue. Subjects were not informed which hand was the refer
ence and were given only 1 s to make a match. Under these conditions,
subjects appeared to match central commands rather than perceived forc
e. Forces were always exaggerated when squeezing an isometric load to
match a compliant load and were too small when matching a compliant lo
ad to an isometric load. The matching errors were largest for the most
compliant spring and increased with the reference force. When the loa
ds were symmetric (e.g., both compliant), errors were small, but subje
cts persistently squeezed slightly harder with the left (nondominant)
hand. Apart from the left-hand bias, the matching results agree quanti
tatively and in detail with the predictions of an equilibrium point mo
del (the a model), in which central commands shift the rest length of
spring-like muscle. The matching data are fit best assuming that the m
uscle's compliance characteristic is an accelerating function, in agre
ement with previous direct measurements of pinch stiffness.