Cerebellar nucleus neurons were recorded in vitro, and dynamic clamping was
used to simulate inhibitory synaptic input from Purkinje cells likely to o
ccur in vivo. Inhibitory input patterns with varying synaptic amplitudes an
d synchronicity were applied to determine how spike rate and spike timing c
an be controlled by inhibition. The excitatory input conductance was held c
onstant to isolate the effect of dynamic inhibitory inputs on spiking. We f
ound that the timing of individual spikes was controlled precisely by short
decreases in the inhibitory conductance that were the consequence of synch
ronization between many inputs. The spike rate of nucleus neurons was contr
olled in a linear way by the rate of inhibitory inputs. The spike rate, how
ever, also depended strongly on the amount of synchronicity present in the
inhibitory inputs. An irregular spike train similar to in vivo data resulte
d from applied synaptic conductances when the conductance was large enough
to overcome intrinsic pacemaker currents. In this situation subthreshold fl
uctuations in membrane potential closely followed the time course of the co
mbined reversal potential of excitation and inhibition. This indicates that
the net synaptic driving force for realistic input levels in vivo may be s
mall and that synaptic input may operate primarily by shunting. The accurat
e temporal control of output spiking by inhibitory input that can be achiev
ed in this way in the deep cerebellar nuclei may be particularly important
to allow fine temporal control of movement via inhibitory output from cereb
ellar cortex.