Spike (Extracellular Action Potential)

The action potential itself is a membrane event inside the neuron; the spike is what its return current leaves in the tissue, which is why the extracellular waveform is smaller, briefer and shaped differently from the intracellular one. It is the strongest transmembrane current a neuron produces, but it lasts under two milliseconds and falls off within tens of micrometers of the soma, so spikes add to the field potential only when many neurons fire together, which they do in the band above about 100 Hz (Buzsáki et al. (2012)).

Amplitude falls steeply with distance. A site within a few tens of micrometers of a cell body sees a large, well-formed waveform; at about a hundred micrometers the spike sinks into the background of other neurons' activity. That fall-off is the detection radius, and it is the number behind two probe decisions: how far apart sites should be, and how many neurons a shank can be expected to yield. Sites closer together than the radius see the same neuron from several positions, which is what lets a tetrode or polytrode layout assign spikes to units; sites farther apart than it see different neurons.

Site properties act on the spike directly. A smaller site with higher impedance is dominated by its nearest neurons and resolves them more sharply, at the cost of a higher noise floor. Spikes are isolated by a band-pass of about 300 Hz to 6 kHz, a filter setting rather than a property of the signal, and are sampled at 20 to 30 kHz. Assigning spikes to neurons is spike sorting; what a sorted neuron is, is a unit.