Representations of subsecond duration-based timing by complex spike synchrony in cerebellar Purkinje neurons

The cerebellum contributes to associative motor learning and sensorimotor coordination in part by tracking subsecond time intervals between behaviorally relevant events, raising the question of how duration, or absolute time, is encoded. Here, we investigated whether information about duration is present in Purkinje cell complex spikes during repetitive sensory stimuli. Crus 1 Purkinje cells expressing the fast calcium indicator GCaMP8f were imaged at high speed (250 fps), allowing detection of complex spike-associated calcium signals from hundreds of Purkinje cell dendrites simultaneously, with 4-ms temporal resolution, in awake head-fixed mice of both sexes. Air puffs were applied to the whisker pad in stimulus trains that varied in the mean and variance of interstimulus intervals (ISIs, 100–900 ms). In responsive cells, the mean probability of complex spike firing increased about fivefold ∼35 ms post-puff, primarily owing to well-timed spiking after the stimulus rather than an increase in spike rate. The maximal response probability, and in some cells also the response latency, varied linearly with ISI. The values of both variables were consistent for each ISI, regardless of the attributes of the stimulus train, suggesting that they carried information about absolute, rather than relative, durations between stimulus pairs. Because each puff evoked only one or zero complex spikes per Purkinje cell, the dependence of spike probability on ISI emerged as a trial-by-trial dependence of the degree of synchronous firing on ISI, suggesting that subsecond absolute timing of somatosensory signals may be represented by complex spike synchrony across populations of Purkinje cells. Significance Statement The cerebellum regulates motor behaviors that require tracking time intervals briefer than one second. Here, we tested how the complex spikes of cerebellar Purkinje cells may encode the interval duration between somatosensory stimuli. We find that at least two attributes of complex spiking vary linearly with the interval between pairs of stimuli, namely, firing likelihood and firing latency. Because of their low spiking probability, however, individual Purkinje neurons cannot accurately report timing information about single stimulus pairs. Nevertheless, the number and latency of complex spikes fired synchronously across the Purkinje cell population changed consistently with the absolute interval between somatosensory stimuli. These data thus demonstrate one way in which neurons may represent time.