Spike sorting
Also: spike classification, single-unit isolation, unit sorting
By neuraspeak editorial · Updated 2026-10-07 · 1 min read
Spike sorting is the computational process of detecting brief electrical pulses (spikes, or action potentials) in a microelectrode recording and assigning each spike to the individual neuron most likely to have produced it, based on the shape of its waveform.
Why is spike sorting necessary?
A single electrode tip in the cortex usually picks up spikes from several nearby neurons at once, mixed with background noise. Each neuron's spike tends to have a characteristic shape and size at a given electrode, so algorithms detect candidate spikes, extract features of their waveforms, and cluster them into groups that correspond to putative single neurons. A classic review described this as a problem of signal detection and classification (Lewicki, Network: Computation in Neural Systems 1998). Modern tools such as Kilosort automate the process for high-density probes, but the task remains hard because recordings drift and nearby neurons' signals overlap (Pachitariu et al., Nature Methods 2024).
Do brain-computer interfaces need spike sorting?
Often not. Many clinical BCIs skip sorting and simply count threshold crossings, every time the voltage on an electrode dips below a set level, regardless of which neuron caused it. A 2019 Neuron study found that neural population dynamics estimated from such multi-unit threshold crossings closely matched those from sorted neurons, which supports this shortcut (Trautmann et al.). Skipping sorting saves computation, avoids errors when waveforms change over time, and still uses electrodes whose signals cannot be cleanly separated into single neurons.
Why it matters for speech BCI
The fastest intracortical speech neuroprostheses rely on unsorted signals. Willett et al. (Nature 2023) decoded speech at 62 words per minute from multi-unit threshold crossings and spike band power rather than sorted single neurons. For speech BCIs, which must run in real time for hours a day over months, the robustness of unsorted features usually matters more than knowing exactly which neuron fired.
Questions
What is a threshold crossing?+
A threshold crossing is a recorded event when an electrode's voltage passes a preset level, usually a multiple of the background noise. It counts spikes from all nearby neurons together, without assigning them to individual cells.
Does spike sorting apply to ECoG or EEG?+
No. ECoG and EEG record summed activity from large populations of neurons and cannot detect individual spikes. Spike sorting applies to penetrating microelectrodes such as the Utah array or Neuropixels probes.
Related: Utah arrayNeural decoderNeural signal driftEEG vs intracortical recording
Sources
- Lewicki, A review of methods for spike sorting: the detection and classification of neural action potentials, Network: Computation in Neural Systems 9 · 1998
- Trautmann et al., Accurate Estimation of Neural Population Dynamics without Spike Sorting, Neuron · 2019
- Pachitariu, Sridhar, Pennington and Stringer, Spike sorting with Kilosort4, Nature Methods · 2024
- Willett et al., A high-performance speech neuroprosthesis, Nature 620, 1031-1036 · 2023
Facts on this page are drawn from the sources listed and were accurate to the best of our knowledge on the update date shown. Company and product names are trademarks of their owners and are used for identification only. Nothing here is medical, investment or legal advice. Companies and individuals can request a correction via the corrections page.