All terms
Common average reference (CAR) is a re-referencing technique that subtracts the mean signal across all (or a subset of) channels from each individual channel, reducing noise that is common across channels.
Why it matters: Many noise sources (movement artifacts, power-line interference, reference electrode drift) affect all channels similarly. CAR removes these shared components, improving signal-to-noise ratio for both spikes and LFPs.
Implementation: For each time sample, compute the mean across all channels, then subtract it from each channel. Can be applied to all channels or restricted to spatial subsets (e.g., per-shank CAR).
Considerations: CAR assumes noise is spatially uniform. If a few channels have large artifacts, median referencing may be more robust than mean referencing.
Guides, downloads, and support docs for Radiens and NeuroNexus hardware.
Open Resources Hub →