Many plasticity mechanisms have been characterized in vitro in circuits important for learning. We are beginning to parse out the dependence of motor learning upon different plasticity mechanisms in the cerebellum and related structures. The most influential model of cerebellum-dependent learning has focused on a single plasticity mechanism, long-term depression (LTD) at parallel fiber to Purkinje cell (pf-Pk) synapses (Albus 1971, Marr 1969), but our studies of the vestibulo-ocular reflex (VOR) suggest that multiple plasticity mechanisms are involved. Applied to motor learning in the VOR, the Marr-Albus model attributes both increases and decreases in VOR gain to pf-Pk LTD (Ito 1982). Here we present two lines of evidence that increases and decreases in VOR gain depend upon different plasticity mechanisms.
First, increases in VOR gain are reversed more readily than decreases in VOR gain. This difference in reversal properties suggests that increases and decreases in VOR gain are mediated by different plasticity mechanisms. Furthermore, the behavioral asymmetry suggests that these plasticity mechanisms reverse each other with unequal efficacy (Boyden & Raymond 2003). The plasticity mechanisms at the pf-Pk synapse (LTD and two forms of LTP) seem to possess such an asymmetric reversal property. Thus we propose a new model in which pf-Pk LTD contributes primarily to increases in VOR gain, whereas pf-Pk LTP contributes primarily to decreases in VOR gain. We are testing this model with mutant mice deficient in either pf-Pk LTD or one of the forms of pf-Pk LTP. Mice lacking Ca+2/CaM-kinase IV (CaMKIV), a molecule required for the late phase of pf-Pk LTD are selectively impaired in retention of an increase in VOR gain. Acquisition of increases and decreases in VOR gain is normal, as is retention of a decrease in VOR gain (Boyden et al 2003). Therefore, long-term memory for an increase in VOR gain relies upon a CaMKIV-dependent process (such as pf-Pk LTD), whereas long-term memory for a decrease in gain does not. Our results suggest that models of cerebellum-dependent motor learning should be revised to consider the role of inverse plasticity mechanisms at pf-Pk synapses.