Local field potential (LFP): definition, origins and uses in organoids

A local field potential (LFP) is the low-frequency component of the extracellular voltage recorded by an electrode in neural tissue. It reflects the summed transmembrane currents of many neurons around the electrode, mostly synaptic, and reveals rhythms, synchrony and network state. In brain organoids, LFP-like oscillations are a key sign of maturing, connected networks.

Key takeaways

• The LFP is the slow part of an extracellular recording; the fast part contains spikes from individual neurons.
• It arises from the overlapping currents of many cells, with synaptic currents a major contributor, as reviewed by Buzsáki, Anastassiou and Koch (2012).
• LFPs, ECoG and EEG are produced by the same physics, recorded at different distances from the tissue.
• Trujillo et al. (2019) showed that human cortical organoids develop increasingly complex oscillatory activity over months.
• For organoid intelligence, the LFP is a population-level readout of network state that helps interpret spikes, stimulation responses and learning.

What is a local field potential? #

When an electrode sits in or against neural tissue, it records a single voltage trace. That trace mixes signals across a wide range of frequencies. The fast, sharp deflections that last about a millisecond are action potentials from neurons very close to the electrode. Underneath them is a slower, rolling signal: the local field potential.

The word “local” matters. An LFP electrode sits inside the tissue (or, for cultures and organoids, directly under it), so it samples a neighborhood of cells rather than an entire brain region. The word “field” refers to the electric field in the extracellular fluid created by currents flowing in and out of many cells at once.

Researchers usually separate the two parts by filtering. A low-pass filter keeps the LFP band, conventionally the frequencies below a few hundred hertz, and a high-pass filter keeps the spike band. The exact cutoffs vary between labs and recording systems, which is one reason good electrophysiology practice reports filter settings with the data.

Where does the LFP signal come from? #

Every active neuron moves ions across its membrane. Current that flows into a cell at one point must flow out somewhere else, so each neuron acts like a tiny current source and sink in the surrounding fluid. An electrode measures the voltage created by all of these sources added together.

Buzsáki, Anastassiou and Koch reviewed the origins of these signals in Nature Reviews Neuroscience (2012). Their key points, in plain terms:

• Synaptic currents are a major contributor. Synaptic inputs are slower than spikes and occur across large areas of dendrite, so they sum well across many cells.
• Other currents contribute too. Spike afterpotentials, calcium spikes, intrinsic membrane oscillations and other processes can all shape the field.
• Geometry matters. Neurons arranged in parallel, such as cortical pyramidal cells with aligned dendrites, produce fields that add up. Randomly oriented cells tend to cancel each other.
• Synchrony matters. Currents that occur at the same time add constructively, so synchronized activity produces larger fields.
• Volume conduction blurs locality. Fields spread through tissue, so an electrode can pick up activity generated some distance away.

This is why an LFP is often described as reflecting the inputs and local processing of a population, while spikes reflect the outputs of individual cells. The relationship is not perfect, and interpreting an LFP always requires care.

How does LFP compare with other neural signals? #

The microelectrode array used for cultures and organoids records both spikes and LFPs from the same electrodes. Calcium imaging offers a complementary optical view, while patch clamp gives the most detailed picture of a single cell.

Common frequency bands #

LFP and EEG researchers often describe rhythms using named frequency bands, such as delta, theta, alpha, beta and gamma, running from slow to fast. The exact band boundaries differ somewhat between fields and species, so it is safer to report actual frequencies than rely on band names alone, especially in organoids, whose rhythms do not necessarily match those of an adult brain.

How are LFPs recorded and analyzed? #

A typical LFP workflow on a cultured network or organoid looks like this:

1. Record broadband voltage from each electrode of an MEA or high-density microelectrode array.
2. Low-pass filter and downsample the signal to keep the LFP band and reduce data volume.
3. Remove artifacts, including line noise and stimulation artifacts that can swamp the slow signal.
4. Compute spectra to see how much power sits at each frequency and how this changes over time.
5. Detect events, such as oscillatory bursts, and relate them to spikes and network bursts.
6. Measure coupling between electrodes, for example coherence or phase relationships, to see how activity spreads.

Analyses often link the two signal types. For example, researchers ask whether spikes tend to occur at a particular phase of an oscillation, or whether population bursts in the spike data line up with large LFP events. Beggs and Plenz described cascades of activity they called neuronal avalanches in The Journal of Neuroscience (2003) using field-potential events recorded on MEAs, which shows how useful the LFP can be for studying collective dynamics.

Why do local field potentials matter for brain organoids? #

A window into network development #

A key question for any brain organoid is whether its neurons form a functional network rather than a set of isolated cells. Spikes alone can show that neurons fire. Field potentials help show that they fire together, in coordinated patterns.

Trujillo et al. reported in Cell Stem Cell (2019) that human cortical organoids recorded on MEAs over months developed oscillatory network events that became more complex as the cultures matured. The authors compared features of this activity with EEG from preterm human infants, framing organoids as a model of early network development. The study is a common reference point for organoid electrophysiology, and our guide on how brain organoids are grown covers the developmental context.

A readout of state in organoid intelligence #

In organoid intelligence, researchers send information into a living network and read out its responses. The LFP helps in several ways:

• Baseline state. It shows whether the network is quiet, oscillating or bursting before a stimulus arrives, which can strongly affect the response.
• Evoked responses. Stimulation often produces a characteristic field response, giving a population-level measure of how the network reacted.
• Change over training. Shifts in rhythm or synchrony across a session can indicate that something in the network has changed, although they do not by themselves prove learning.
• Feature extraction. In reservoir computing, any reliable, input-dependent signal can serve as a feature for a trained readout. Cai et al. (Nature Electronics, 2023) used recorded organoid activity in this way in their Brainoware study.

Smirnova et al. (Frontiers in Science, 2023) set out a roadmap for organoid intelligence in which richer, multi-scale readouts of organoid activity are a central technical need. LFPs are part of that toolkit, alongside spikes and optical methods.

What are the limitations of LFP recordings? #

• Mixed sources. An LFP is a sum of many currents, so it rarely maps cleanly onto one cell type or mechanism.
• Uncertain locality. Volume conduction means some signal may come from farther away than the electrode’s immediate neighborhood.
• Geometry effects. Organoids lack the regular layered structure of adult cortex, which can make fields weaker or harder to interpret than in intact brain.
• Reference choice. The recorded voltage depends on the reference electrode, and different choices can change apparent results.
• Artifacts. Stimulation, movement of the medium and electrical noise all fall in or near the LFP band.
• Surface access. A planar array mainly samples the parts of a three-dimensional organoid that touch the electrodes.

These limitations are manageable with careful design, consistent processing and transparent reporting. Shared formats such as Neurodata Without Borders make it possible to store raw broadband data alongside filtered LFPs, so others can re-analyze them with different choices.

How does LFP fit into closed-loop experiments? #

In closed-loop stimulation, software reads neural activity and decides on the next stimulus in real time. Most closed-loop systems for cultured neurons key on spikes, because they are fast and well localized. LFP features can still play a role, for example by timing a stimulus to a particular phase of an ongoing oscillation, or by pausing stimulation when the network enters a large synchronized burst. The DishBrain work, covered in our article on how neurons learn in a dish, shows how closely recording, decoding and feedback are tied together in these loops.

Treating LFP and spike streams as first-class, versioned data is part of what we are building at Organoid Programming, so that every analysis choice can be traced and repeated. For a hardware view of how these signals are captured, see microelectrode arrays explained. The local field potential remains one of the simplest and most informative ways to ask whether a living network is behaving like a network.

Frequently asked questions

What is a local field potential in simple terms?

A local field potential is the slow, rolling voltage signal an electrode picks up when many neurons near it are active together. Instead of showing individual spikes, it shows the combined electrical activity of a neighborhood of cells, mostly from synaptic currents. Scientists use it to see rhythms, synchrony and the overall state of a neural network.

What is the difference between LFP and spikes?

Spikes are fast, brief signals from individual neurons firing action potentials, typically isolated by high-pass filtering an extracellular recording. The LFP is the slower, low-frequency part of the same recording and reflects the summed currents of many cells, especially synaptic inputs. Spikes mainly tell you about output from nearby neurons, while the LFP tells you more about the inputs and coordinated activity of a local population.

How is LFP different from EEG?

Both are extracellular field potentials produced by the same physical processes, but they are recorded at different distances. The LFP is measured with electrodes inside or directly against neural tissue and reflects activity from a local population. EEG is recorded on the scalp, through skull and skin, so it captures a blurred sum of activity from large areas of cortex.

Do brain organoids produce local field potentials?

Yes. Trujillo et al. reported in Cell Stem Cell (2019) that human cortical organoids recorded on microelectrode arrays developed oscillatory network activity over months of culture, with patterns that grew more complex as the organoids matured. These low-frequency field signals are one of the main ways researchers track how organoid networks develop.

Why are local field potentials useful for organoid intelligence?

LFPs give a population-level view of network state that complements spike recordings. In organoid intelligence experiments they can show whether a network is synchronized, bursting or quiet before a stimulus arrives, how it responds afterwards, and how its rhythms change over training. That context helps researchers interpret spikes and design better encoding, decoding and feedback strategies.

Sources

1. Buzsáki, Anastassiou & Koch (2012). The origin of extracellular fields and currents: EEG, ECoG, LFP and spikes. Nature Reviews Neuroscience.
2. Trujillo et al. (2019). Complex oscillatory waves emerging from cortical organoids model early human brain network development. Cell Stem Cell.
3. Cai et al. (2023). Brain organoid reservoir computing for artificial intelligence. Nature Electronics.
4. Beggs & Plenz (2003). Neuronal avalanches in neocortical circuits. The Journal of Neuroscience.
5. Smirnova et al. (2023). Organoid intelligence (OI): the new frontier in biocomputing and intelligence-in-a-dish. Frontiers in Science.