Spike Train

A spike train is the time-ordered sequence of one unit's spike times, stored as timestamps at a resolution of about 30 µs to 1 ms; it is the output of spike sorting and the input to every single-neuron measure. Firing rate is spikes per second over a window. The inter-spike interval is the time between consecutive spikes, and its histogram shows regularity, bursting — a cluster of intervals under about 10 ms — and the refractory period, where intervals under about 1 to 2 ms are absent for a well-isolated unit and present when two neurons have been merged. The autocorrelogram counts spikes at each lag from every spike in the train, typically over ±50 to 500 ms in 1 ms bins: a gap at zero lag confirms isolation, periodic peaks show oscillatory firing, and a broad peak at short lags shows bursting.

What the electrode adds is provenance. A spike train is only as trustworthy as the unit it came from, so the same interval and correlogram measures that describe a neuron's firing are also the quality checks on its sorting, and a train whose refractory gap fills in over a session is a unit that has drifted. Event-aligned views of a train are the raster and PSTH. In Radiens the spike train is sorted in real time in Allego and kept with its provenance through Videre, so the train a paper reports is traceable to the recording it came from.