Local field potentials, or LFPs, are the low-frequency electrical signals recorded from inside brain tissue that reflect the collective activity of thousands of neurons in a region. When brain cells communicate, they generate tiny currents that flow through the fluid and tissue surrounding them. An electrode placed in the brain picks up the summed effect of all those currents, producing a smoothly fluctuating voltage trace that neuroscientists call the local field potential. LFPs sit in a useful middle ground between the activity of single neurons and the broad electrical patterns captured by scalp electrodes, and that position makes them one of the most informative signals in modern neuroscience.
Where the Signal Comes From
Every time one neuron sends a chemical message to another across a synapse, small currents flow through the receiving cell’s membrane. These synaptic transmembrane currents are the largest contributor to the LFP signal. But they are not the only contributor. Sodium and calcium spikes, currents flowing through ion channels, and intrinsic oscillations in the membrane itself all shape the extracellular field.1Nature Reviews Neuroscience. The origin of extracellular fields and currents — EEG, ECoG, LFP and spikes Think of it like standing in a crowd: the hum you hear is mostly people talking, but footsteps, coughs, and rustling all blend in.
The key point is that LFPs do not simply mirror the firing of individual neurons. A neuron can receive thousands of synaptic inputs without itself producing a spike, and each of those inputs contributes current to the surrounding tissue. As a result, the LFP captures a lot of the “behind the scenes” processing: the inputs arriving at a population of cells, the subthreshold fluctuations that push them closer to or further from firing, and the inhibitory currents that keep them in check. That is precisely why LFPs contain information that single-neuron recordings alone would miss.
How Far Does the Signal Travel
A common assumption is that LFPs reflect only the activity of cells in the electrode’s immediate neighborhood, perhaps a few hundred micrometers away. Reality is more complicated. Direct measurements have shown that LFPs can spread passively to sites more than a centimeter from their source, and this spread appears to hold regardless of the frequency content of the signal.2Neuron. Local Field Potentials: What They Are and Why They Matter In one set of recordings from auditory cortex, a prominent component of the LFP remained visible all the way up to the surface of the brain, roughly 18 mm above its origin.
This means that an LFP recording is not purely “local.” It is a mixture of potentials generated nearby and potentials that have traveled from more distant sources through a process called volume conduction. Computational modeling supports this picture and adds nuance: the spatial reach of the LFP varies with frequency, with the level of correlation among nearby inputs, and with the shape and arrangement of the neurons generating the currents.3PLoS Computational Biology. Frequency Dependence of Signal Power and Spatial Reach of the Local Field Potential There is no single fixed radius you can draw around an electrode and call it the LFP’s territory. The answer depends on the brain region, the cell types involved, and even the kind of activity happening at the time.
Researchers can partially address this ambiguity by computing what is called the current source density, which mathematically strips away volume-conducted signals to reveal where current is actually entering or leaving the tissue. Intracranial current source density analysis can resolve activity at a level finer than individual cortical layers, revealing patterns of sources and sinks that correspond to specific physiological events.4PubMed Central. Generator localization by current source density (CSD): implications of volume conduction and field closure at intracranial and scalp resolutions A detailed model of mouse visual cortex demonstrated something striking: you can adjust synaptic weights and change firing rates without much effect on the current source density pattern, and conversely you can rearrange where synapses land on the dendrites and change the current source density without much effect on firing rates.5eLife. Uncovering circuit mechanisms of current sinks and sources with biophysical simulations of primary visual cortex The LFP and the spiking of neurons are related but not interchangeable. They contain partly independent information about what circuits are doing.
What the Different Frequencies Mean
One of the richest features of LFPs is that they contain multiple rhythmic components happening simultaneously, each associated with different aspects of brain function. You can decompose an LFP recording into frequency bands much the way you might separate the bass, midrange, and treble in a music signal. The major bands carry distinct kinds of information.
Low-frequency fluctuations, below about 12 Hz, tend to track sensory stimuli reliably even over very short time windows. The energy of a single cycle or just a few cycles of a slow oscillation can tell you what stimulus was presented. Faster oscillations above roughly 50 Hz behave differently: their energy carries information about the stimulus only when averaged over hundreds of milliseconds and many cycles. Crucially, the information in different LFP bands is largely independent of each other and of spiking activity, suggesting that each band captures a different dimension of what the brain is doing.6PubMed Central. Sensory information in local field potentials and spikes from visual and auditory cortices: time scales and frequency bands
Gamma-band activity, peaking roughly between 30 and 100 Hz, has attracted particular attention.7PubMed Central. Gamma rhythms in the visual cortex: functions and mechanisms In visual cortex, gamma oscillations are strongly driven by stimulus features and modulated by attention. Interestingly, directing attention toward a stimulus actually decreases gamma power and the coherence between spikes and the gamma-band LFP in the primary visual cortex, which may reflect a reduction in inhibitory drive rather than an enhancement of processing.8PubMed Central. Attention reduces stimulus-driven gamma frequency oscillations and spike field coherence in V1 That finding challenges the once-popular idea that gamma synchronization is the brain’s mechanism for binding features into a unified percept. The story is more complicated than early enthusiasm suggested.
In visual cortex recordings in primates, different LFP frequency bands show diverse tuning to stimulus properties. Higher frequencies above about 140 Hz track the overall population firing rate, lower frequencies up to around 40 Hz respond transiently after a stimulus appears, and a narrow gamma band between roughly 50 and 80 Hz behaves differently from both, showing weak correlation with either.9PubMed Central. Distinct frequency bands in the local field potential are differently tuned to stimulus drift rate Each band seems to convey different attributes of the underlying neural activity, making the LFP a surprisingly multi-dimensional readout of a single brain region.
When Rhythms Talk to Each Other
Brain oscillations do not operate in isolation. One of the more exciting findings in recent years is that slow rhythms can modulate the amplitude of faster rhythms, a phenomenon called phase-amplitude coupling. Imagine a slow wave rolling through a brain region: at the peak of each cycle, bursts of faster activity ride on top of it. This is not a metaphor. It is a measurable, reproducible pattern in LFP recordings.
In rats performing a navigation and decision-making task, the amplitudes of multiple high-frequency oscillations were dynamically modulated by the phase of theta-band oscillations, both within the hippocampus and the striatum and between these two structures. The modulation patterns were not static; they shifted in task-dependent ways, particularly during decision-making moments.10PubMed Central. Dynamic cross-frequency couplings of local field potential oscillations in rat striatum and hippocampus during performance of a T-maze task The implication is that cross-frequency coupling could be a general mechanism the brain uses to coordinate computations across different timescales and different regions.
Analyzing phase-amplitude coupling requires careful signal processing, including preprocessing the LFP, decomposing it into its time-frequency components, and computing the coupling strength between specific frequency pairs.11PubMed Central. Protocol for phase-amplitude coupling analysis in local field potentials from macaque monkeys to investigate neural oscillation dynamics The method has become a standard tool in systems neuroscience, applied to everything from sensory processing to memory consolidation.
LFPs, Memory, and Movement
Some of the most compelling evidence for the functional importance of LFPs comes from studies of memory replay in the hippocampus. During sharp wave ripples, brief high-frequency bursts that occur during quiet wakefulness and sleep, the hippocampus replays sequences of activity that correspond to recently experienced spatial trajectories. Multi-site LFP amplitudes in the roughly 150 to 200 Hz band reliably reflect which constellation of place cells is active during these replays, and spatiotemporal patterns in the LFP remain consistent between replay events that activate the same cell ensembles.12PubMed Central. Local Field Potentials Encode Place Cell Ensemble Activation during Hippocampal Sharp Wave Ripples In other words, the LFP alone can tell you which memory is being reactivated, without needing to record from individual neurons.
In the motor system, LFP oscillations in the basal ganglia have become central to understanding movement disorders. The basal ganglia naturally exhibit rhythmic activity, but in Parkinson’s disease, when dopamine levels drop, certain oscillatory patterns become exaggerated. Some cortical rhythms appear to penetrate into the basal ganglia while others are transformed or blocked, and the balance of these interactions shifts in the disease state.13PubMed Central. Oscillations and the basal ganglia: motor control and beyond These pathological oscillations are not just a side effect of the disease; they appear to actively interfere with normal motor function.
Even in healthy motor cortex, LFP signals carry information about more than just movement planning. In monkeys performing a task where reward size varied, the expectation of a larger reward influenced alpha-band (8 to 14 Hz) power, the coupling between alpha and gamma rhythms, spike-field coherence, and firing rates in the primary motor cortex.14eNeuro. Reward Expectation Modulates Local Field Potentials, Spiking Activity and Spike-Field Coherence in the Primary Motor Cortex Motor cortex, it turns out, is not just a movement command center; its LFP activity reflects motivational context too.
Clinical and Brain-Machine Interface Applications
The practical significance of LFPs extends well beyond basic science. In deep brain stimulation for Parkinson’s disease, electrodes implanted in the basal ganglia deliver electrical pulses to suppress pathological activity. Conventional systems deliver stimulation at a fixed rate regardless of the patient’s current state. Adaptive deep brain stimulation aims to improve on this by using LFPs recorded from the same electrodes as a feedback signal, adjusting stimulation in real time based on the oscillatory biomarkers of the patient’s condition.15PubMed. Adaptive deep brain stimulation (aDBS) controlled by local field potential oscillations Because the pathological beta-band oscillations in Parkinson’s rise and fall with symptom severity, they provide a natural control variable for a closed-loop system.
In epilepsy, LFP recordings from within the seizure onset zone reveal distinct patterns that precede seizures detectable on standard intracranial EEG. In cases with hypersynchronous seizure onset, fast ripples with progressively increasing power accompanied epileptiform discharges during the transition to seizure.16PubMed Central. Ictal onset patterns of local field potentials, high frequency oscillations, and unit activity in human mesial temporal lobe epilepsy Detecting these signatures early could eventually allow closed-loop devices to intervene before a seizure fully develops.
Brain-machine interfaces represent another area where LFPs are proving their worth. Most high-performance interfaces rely on recordings of individual neuron spikes, but spike signals tend to degrade over months as recording electrodes develop scar tissue. LFPs are far more stable over time. In one study, monkeys used an LFP-driven brain-machine interface to control a computer cursor with high performance that remained stable or even improved over nearly 12 months, with no need to retrain the decoder.17Journal of Neural Engineering. Long term, stable brain machine interface performance using local field potentials and multiunit spikes When spike signals are available, combining them with LFPs can boost performance; when spikes fade, LFPs can serve as a reliable fallback.18Journal of Neural Engineering. A high performing brain–machine interface driven by low-frequency local field potentials alone and together with spikes
The Technical Challenge of Clean Recordings
Getting meaningful information from LFPs is not as simple as placing an electrode and reading out the voltage. One persistent headache is spike contamination. Because the same electrode that picks up slow LFP fluctuations also records the sharp voltage transients of action potentials from nearby neurons, simply low-pass filtering the raw signal can bleed spike waveforms into the LFP band and create artifactual correlations between spikes and the field potential. Simulations have confirmed that these artificial correlations exert a powerful influence on popular measures of spike-LFP synchronization.19PubMed. Removal of spurious correlations between spikes and local field potentials After applying dedicated spike-removal algorithms, many recordings still show genuine spike-LFP correlations, confirming that the relationship is real but also that careful processing is essential to distinguish real coupling from artifact.
Multiple approaches exist for removing spike contamination. Some use linear filtering to subtract features correlated with spike events from the LFP trace.20PubMed Central. Decoupling action potential bias from cortical local field potentials More recent methods decompose individual action potentials into their frequency components and remove each component adaptively, accounting for the variable shape and timing of spike artifacts.21PubMed Central. Adaptive Spike-Artifact Removal from Local Field Potentials Uncovers Prominent Beta and Gamma Band Neuronal Synchronization The choice of artifact removal method can qualitatively change results, especially in the beta and gamma bands where spike energy overlaps most with the LFP frequencies of interest.
Spike-field coherence, the measure of how tightly individual neuron firing locks to a particular phase of the LFP oscillation, is one of the most widely used metrics in the field. It quantifies, for each frequency, how consistently spikes occur at the same point in the oscillation cycle, producing a value between zero and one.22iScience. Low-Frequency Spike-Field Coherence Is a Fingerprint of Periodicity Coding in the Auditory Cortex When spike-field coherence is high, it implies that the neuron’s firing is organized by the rhythm, not happening randomly with respect to it. Getting this metric right depends entirely on having clean separation between spikes and LFP.
Modeling LFPs Computationally
Interpreting LFPs is hard enough from recordings alone. Computational models have become an important complement. The challenge is that realistic models of LFP generation require multicompartment neuron models that simulate currents flowing through detailed dendritic trees, while large-scale network simulations typically use simplified “point neurons” that lack spatial structure. A hybrid approach bridges this gap: researchers run a network simulation with efficient point neurons to capture the dynamics and then map those dynamics onto a population of biophysically detailed neuron models to predict what the LFP would look like.23PubMed Central. Hybrid Scheme for Modeling Local Field Potentials from Point-Neuron Networks This kind of modeling has helped clarify which features of the LFP depend on the network’s firing patterns and which depend on the physical arrangement of synapses and dendrites.
LFPs Beyond Wakefulness
LFP patterns change dramatically with brain state. Under isoflurane anesthesia in mice, the peak frequency of the theta rhythm in the hippocampus drops from about 8 Hz during wakefulness to roughly 6 Hz during induction, then continues to decline through deeper stages of sedation. During recovery, delta activity speeds up first after the animal regains its righting reflex, and theta rhythms gradually return to normal only once the animal resumes active locomotion.24Frontiers in Cellular Neuroscience. Electrophysiological activity pattern of mouse hippocampal CA1 and dentate gyrus under isoflurane anesthesia Tracking these frequency shifts in real time could help anesthesiologists gauge depth of anesthesia more precisely than current methods allow.
The non-canonical contributors to LFP signals add another layer of complexity. Astrocytes, the star-shaped glial cells long considered mere support staff, generate their own slow electrical currents as they take up glutamate released at synapses and buffer excess potassium ions left behind by spiking neurons. These astrocytic signals are detectable in extracellular recordings and overlap in frequency with the slowest LFP components.25PubMed Central. Neuron-glia cross talk revealed in reverberating networks by simultaneous extracellular recording of spikes and astrocytes’ glutamate transporter and K+ currents How much astrocytic activity shapes LFPs in the intact brain remains an open question, but it underscores that the LFP is not exclusively a neuronal signal.
Brain Rhythms Across the Animal Kingdom
One of the more surprising facts about brain oscillations is how conserved they are across species. Despite the several-thousand-fold increase in brain volume over the course of mammalian evolution, the hierarchy of brain rhythms has remained remarkably preserved, allowing communication within and across neuronal networks at approximately the same speed regardless of brain size.26PubMed Central. Scaling brain size, keeping timing: evolutionary preservation of brain rhythms A mouse hippocampus and a human hippocampus produce theta rhythms in roughly the same frequency range, even though the human version is vastly larger.
Extending beyond mammals, researchers have recorded LFP-like signals in reptiles during sleep. In the bearded dragon, two distinct electrophysiological sleep states were identified, one of which features slow waves reminiscent of mammalian slow-wave sleep. But in the Argentine tegu, a different reptile, the sleep-state patterns looked quite different: one state lacked the slow negative waves seen in the dragon, while the other was dominated by an oscillation around 15 Hz with no clear mammalian analog.27PLoS Biology. Partial homologies between sleep states in lizards, mammals, and birds suggest a complex evolution of sleep states in amniotes These cross-species comparisons suggest that organized oscillatory activity in the brain predates the mammalian lineage, but the specific patterns have diverged in ways that mirror each lineage’s behavioral demands. The LFP, far from being a mere laboratory tool, is a window into one of the most fundamental features of how nervous systems organize themselves.