The motor system may use internal predictive models of the motor appar
atus to achieve better control than would be possible by negative feed
back. Several theories have proposed that the cerebellum may form thes
e predictive representations. In this article, we review these theorie
s and try to unify them by reference to an engineering control model k
nown as a Smith Predictor. We suggest that the cerebellum forms two ty
pes of internal model. One model is a forward predictive model of the
motor apparatus (e.g., limb and muscle), providing a rapid prediction
of the sensory consequences of each movement. The second model is of t
he time delays in the control loop (due to receptor and effector delay
s, axonal conductances, and cognitive processing delays). This model d
elays a copy of the rapid prediction so that it can be compared in tem
poral register with actual sensory feedback from the movement. The res
ult of this comparison is used both to correct for errors in performan
ce and as a training signal to learn the first model. We discuss evide
nce that the cerebellum could form both of these models and suggest th
at the cerebellum may hold at least two separate Smith Predictors. One
, in the lateral cerebellum, would predict the movement outcome in vis
ual, egocentric, or peripersonal coordinates. Another, in the intermed
iate cerebellum, would predict the consequences in motor coordinates.
Generalization of the Smith Predictor theory is discussed in light of
cerebellar involvement in nonmotor control systems, including autonomi
c functions and cognition.