Motor imagery typically involves an experience as of moving a body par
t. Recent studies reveal close parallels between the constraints on mo
tor imagery and those on actual motor performance. How are these paral
lels to be explained? We advance a simulative theory of motor imagery,
modeled on the idea that we predict and explain the decisions of othe
rs by simulating their decision-making processes. By proposing that mo
tor imagery is essentially off-line motor action, we explain the tende
ncy of motor imagery to mimic motor performance. We close by arguing t
hat a simulative theory of motor imagery gives (modest) support to and
illumination of the simulative theory of decision-prediction.