Time series analysis of neuronal signals recorded in the cerebellum of trained monkeys

Recordings from cerebellar neurones have been made in monkeys trained to track a target moving sinusoidally at 0·33 Hz. The neuronal point even time series was first low pass filtered at 0·33 Hz to compare with the target and monkey movements. Short lengths of data were then subjected to Fourier and autoregressive spectral analysis in order to characterize higher frequencies in the neuronal signal. Stable components were identified at around 0·33 Hz (when the monkey was tracking well), 1·0 Hz and 8·5 Hz. The variation of many of the intermediate frequencies was suggestive of entrainment interactions between the basic components. The power at 1·0 Hz may result from the operation of a visuomoter control loop, whilst 8·5 Hz may characterize proprioceptive control.