What Are Local Field Potentials?

Local field potentials are the low-frequency component of the extracellular voltage recorded by a microelectrode placed within neural tissue. They are obtained by low-pass filtering the raw extracellular signal, conventionally below about 300 Hz, which separates them from the high-frequency band containing action potentials. Where a spike reports the output of a single neuron near the electrode tip, a local field potential reflects the summed transmembrane currents of a population of cells within a surrounding volume of tissue, weighted by distance and by the geometry of the current sources. This makes the signal a measure of collective network activity rather than of individual firing.

The dominant contribution to local field potentials comes from synaptic currents in the dendrites of pyramidal neurons, whose elongated and spatially aligned morphology produces open-field current dipoles that do not cancel at a distance. Slower processes including calcium spikes, afterhyperpolarizations, and glial membrane currents also contribute. Because the tissue acts as a volume conductor, the measured potential at any point is a spatially filtered mixture, and interpreting it requires forward models based on current source density theory rather than a direct reading of nearby activity.

Biophysical Origin and Spatial Reach

The spatial reach of a local field potential has been contested since the first cortical recordings, with published estimates ranging from a few hundred micrometers to several millimeters. The answer depends on the spatial correlation of synaptic input rather than on tissue conductivity alone: correlated input across a population extends the reach considerably, while uncorrelated input keeps it local. Modeling and experimental work on the information content of local field potentials shows that different frequency bands carry partly independent information about sensory stimuli, with low-frequency power tracking slow network state and gamma-band power tracking local processing. Laminar electrode arrays combined with current source density analysis are used to localize sinks and sources across cortical layers and recover the underlying synaptic activation pattern.

Recording and Signal Processing

Local field potentials are recorded with metal microelectrodes, silicon probes, and high-density CMOS arrays, and the same electrode usually yields both spike and field bands after separate filtering. Signal quality depends heavily on the reference scheme. Ground-referenced recordings pick up distant volume-conducted activity and common-mode noise, and work on differential recording of local field potentials shows that bipolar derivations improve signal-to-noise ratio and give a more faithful estimate of functional connectivity between sites. Standard analyses include multitaper spectral estimation, time-frequency decomposition, phase-amplitude coupling, coherence between electrode pairs, and Granger causality. Machine learning has added a further use: a study in Scientific Reports on inferring spiking activity from local field potentials demonstrates that deep networks can recover population firing rates from the field signal alone, which matters for implants where recording spikes long term is difficult.

Clinical and Neural Interface Use

Local field potentials are attractive for chronic implants because they are far more stable over months and years than single-unit recordings, which degrade as the electrode-tissue interface changes. Deep brain stimulation devices now sense field potentials from their own leads, and beta-band power in the subthalamic nucleus serves as a feedback signal for adaptive stimulation in Parkinson's disease. Brain-computer interfaces decode movement intention from field potential features in motor cortex, and epilepsy monitoring uses high-frequency oscillations riding on the field signal to help localize seizure onset zones.

Applications

Local field potential recording and analysis are used in fields including:

• Systems neuroscience studies of sensory coding, attention, and memory
• Closed-loop deep brain stimulation for movement disorders
• Brain-computer interfaces and neuroprosthetic control
• Epilepsy source localization and seizure prediction
• Preclinical pharmacology and neurotoxicology screening
• Low-power biomedical circuit design for implantable neural recording