Every clinical neurophysiology recording — scalp EEG, surface NCS, needle EMG, evoked potentials — captures electrical activity that has been spatially smoothed, attenuated, and sometimes inverted by the tissues between the source and the electrode. This is volume conduction: the body acts as a complex resistor-capacitor network through which current flows from active generators to recording electrodes. Understanding volume conduction explains why a focal cortical seizure appears widespread on scalp EEG, why a distally recorded sensory response can be larger than one recorded proximally, why EEG polarity can reverse across the head, and why bipolar vs referential montages give different pictures of the same source. This page covers the recording geometry, source localization principles, and the technical setup that determines what we see.

Volume Conduction Basics

Body tissues (skin, fat, muscle, bone, CSF, brain) all conduct current with different resistivities.
An electrical source in the brain or peripheral nerve creates a 3D field that spreads through the entire conducting medium.
The voltage measured at any electrode is the algebraic sum of contributions from all active sources in the conducting volume.
Distance, source strength, dipole orientation, and tissue resistivity determine each contribution.

Implication 1: Scalp EEG Captures Distributed Cortical Activity

A single neuron contributes essentially zero detectable signal at the scalp.
Several square centimeters of synchronously active cortex are required to produce a recordable scalp deflection (typically 6 cm² minimum).
A focal cortical epileptic spike often appears over multiple electrodes because the field spreads through the skull and scalp.
Conversely, a brief, sharply-contoured artifact at the scalp may have no cortical source at all.

Implication 2: NCS Recordings Pick Up More Than the Target Muscle

The recording electrode over abductor pollicis brevis (median motor) sees not only APB but also nearby thenar muscles, the volume-conducted signal from forearm flexors, and even distant proximal arm activity.
This is why median CMAP can be recorded even with severe carpal tunnel — surrounding muscles via volume conduction give a small response.
This is also why “co-contraction” artifact (the patient activating other muscles) contaminates the recording.

Implication 3: Polarity Depends on Source Orientation

A cortical dipole with its negative pole upward produces negative deflection at electrodes directly above and positive deflection at electrodes 180° away.
The “phase reversal” seen across an epileptiform focus is a direct consequence: the focal source acts as a dipole, with electrodes on opposite sides recording opposite polarities.
Phase reversal in a bipolar montage localizes the source between the electrodes showing maximum opposite polarity.

Recording Geometry

Active and Reference Electrodes

Every voltage measurement is a difference between two points; there is no “absolute” voltage.
The “active” electrode is typically placed over the area of interest; the “reference” is at a location intended to be inactive.
In practice, no reference is truly inactive — every electrode picks up some signal — so the recorded waveform is the active source minus whatever is at the reference.

Monopolar (Referential) Recording

Each active electrode is compared to a single common reference (e.g., averaged ears, mastoid, Cz).
Shows amplitude and polarity of each region in absolute terms.
Good for visualizing widespread activity, comparing across regions, and identifying sources.
Sensitive to reference contamination — if the reference is active, every channel shows that activity.

Bipolar Recording

Each channel is the difference between two adjacent active electrodes (e.g., F7-T3, T3-T5).
Cancels out activity common to both electrodes — emphasizes local differences.
Excellent for localizing focal sources: phase reversal pinpoints the maximum.
Insensitive to widespread activity (common-mode rejection cancels it).
Standard chains: longitudinal bipolar (anterior-to-posterior “double banana”) and transverse bipolar.

Average Reference

Each electrode compared to the mean of all electrodes.
Works well when most electrodes are inactive (e.g., focal abnormality).
Fails when activity is widespread (the “reference” becomes contaminated by the activity itself).

Laplacian (Source Derivation)

Each electrode compared to the weighted average of its immediate neighbors.
Emphasizes very focal sources, suppresses distant activity.
Useful for localizing IEDs and identifying true focal vs widespread sources.

The 10-20 System

Standard scalp electrode positions based on percentages of head dimensions (nasion-inion, preauricular-preauricular).
Letters indicate region: F=frontal, T=temporal, C=central, P=parietal, O=occipital, Fp=frontopolar.
Numbers indicate side: odd=left, even=right, z (zero)=midline.
Standard 21 electrodes give ~6 cm spacing.
10-10 system (extended) doubles electrode density for source localization.
10-5 system increases further for research and intracranial cortical mapping correlation.

Why 10-20 Doesn’t Localize Like MRI

Skull and scalp spatially smooth the cortical signal — a 1 cm² source spreads to ~6 cm at the scalp.
Adjacent electrodes are highly correlated, especially for distant or widespread sources.
Two physically separated cortical sources can produce identical scalp patterns (“inverse problem” — multiple cortical configurations explain the same scalp data).
High-density EEG (64–256 electrodes) plus source modeling can localize within 1–2 cm in good conditions; routine 21-electrode EEG localizes to a lobe or sublobar region only.

NCS Recording Geometry

Belly-Tendon Montage (Motor NCS)

Active electrode over the motor point of the target muscle (where end-plates are concentrated).
Reference over the muscle’s distal tendon (electrically inactive).
Produces a biphasic CMAP with a clean initial negative deflection — the takeoff used to measure latency.
Misplacement (active not over motor point) produces an initial positive deflection — indicates the recording electrode is not in optimal position.

Antidromic vs Orthodromic Sensory NCS

Antidromic: stimulate proximal nerve, record distally over the digital sensory fibers (e.g., stimulate median at wrist, record over index finger). Larger amplitude, less contamination, technically simpler.
Orthodromic: stimulate digital sensory fibers distally, record over the nerve trunk proximally. Smaller signal, more contamination from motor fibers, but anatomically natural direction.
Antidromic is the most common clinical approach for sensory NCS.

Stimulation Geometry

Cathode (negative pole) placed distal to anode for orthodromic motor responses traveling proximal-to-distal.
Reversing polarity reduces effective stimulation by ~20% (anodal block).
Inter-electrode distance affects stimulus duration and spread.

Common Mode Rejection

Differential amplifiers reject signals common to both inputs.
This is how 60-Hz line noise is suppressed: it appears on both active and reference electrodes equally.
Common-mode rejection ratio (CMRR) for clinical amplifiers: typically >100 dB.
If electrode impedances differ greatly between active and reference, CMRR degrades and 60-Hz appears prominently.
Practical implication: always check that all electrode impedances are below 5 kΩ and roughly matched. Unequal impedances are the most common cause of “noisy” recordings.

Filters and Sampling

Low-frequency (high-pass) filter: removes drift below the cutoff. EEG typically 1 Hz (or 0.5 Hz for slow-wave analysis); NCS sensory typically 20 Hz; NCS motor typically 10 Hz; EMG typically 20–30 Hz.
High-frequency (low-pass) filter: removes noise and muscle artifact above the cutoff. EEG typically 70 Hz; NCS typically 2–10 kHz; EMG typically 10 kHz.
Notch filter: targets a specific frequency (60 Hz US, 50 Hz Europe). Use sparingly — can attenuate real biology in that band.
Sampling rate: at least twice the highest frequency of interest (Nyquist theorem). Clinical EEG: 256–512 Hz minimum; HFO analysis: 2000 Hz or higher.

Filter Pitfalls

Aggressive low-pass filter makes sharp transients look like slow waves.
Aggressive high-pass filter makes slow waves look smaller and faster.
Notch filter at 60 Hz can produce ringing artifact (“Gibbs phenomenon”) around true sharp transients.
Wrong filter setting is the most common technical error in over-reading: a “sharp wave” that disappears when filter is widened is not pathologic.
Always look at the raw signal at multiple filter settings before interpreting borderline findings.

Common Artifacts and Their Sources

EEG Artifacts

60 Hz line noise: from poor grounding or unequal electrode impedances. Look for sinusoidal activity at exactly 60 Hz across multiple channels.
Electrode pop: a sharp brief deflection limited to one channel; abrupt baseline shift. Distinguishes from biological discharge by being monopolar (only one channel) and not following any neuroanatomic pattern.
Eye movement: large slow deflections, frontal predominance, lateral channel asymmetry for horizontal saccades (LFp7 vs LFp8).
Muscle: high-frequency (>30 Hz) bursts of variable amplitude, concentrated over frontal and temporal regions.
ECG: rhythmic at heart rate, often visible at the vertex, ipsilateral mastoid, or ear electrodes; large in obese patients.
Sweat: slow undulating waves, frequency below 1 Hz, especially in warm rooms.
Movement: irregular slow waves correlating with patient’s body movement.
Ventilator: rhythmic artifact at ventilator rate.
IV pump / infusion artifact: rhythmic at pump cycle frequency.

NCS/EMG Artifacts

Stimulus artifact: large brief deflection at stimulus onset; contaminates short-latency responses.
Volume-conducted response from adjacent muscle: smaller, often inverted polarity, slightly different timing.
Anodal block: stimulator polarity reversed; response disappears or shrinks.
Cold limb: apparent slowing of conduction; warm to 32°C distally.
Co-contraction: voluntary contraction of antagonist during stimulation; produces “noise” superimposed on the response.
Needle motion artifact (EMG): high-frequency burst as needle is moved.

Practical Setup Principles

Skin preparation: clean with abrasive paste or alcohol; reduces impedance.
Electrode impedance: aim for <5 kΩ; check all channels before recording.
Match impedances: very unequal impedances destroy common-mode rejection.
Ground electrode: required for safety and noise suppression; place between active and stimulator if possible.
Patient temperature: warm cold limbs to 32°C distally for NCS; cooler temperature falsely slows conduction.
Patient cooperation: relaxed patient is essential for NCS, EMG, and EEG; sleep deprivation increases EEG yield for IEDs.
Documented technique: every report should specify electrode placements, filter settings, stimulator settings, activation procedures, and recording time.

🔍 Did You Know?

The “inverse problem” in EEG — given a scalp voltage pattern, what cortical sources produced it? — is mathematically ill-posed: infinitely many cortical configurations can produce the same scalp recording. This is a fundamental limitation of scalp EEG that no improvement in electronics can solve. Realistic head models, high-density electrode arrays (256 channels), and physiologic constraints (e.g., assumption that sources are dipolar and in cortex) reduce the solution space, and modern source localization methods (LORETA, sLORETA, beamforming) can resolve sources to within 1–2 cm under ideal conditions. But the routine 21-electrode EEG used in most clinics localizes to a lobe or sublobar region only — and only when the source is convex-surface cortex with an outward radial component. Sources in mesial temporal lobe, deep frontal lobe, or inferior temporal gyrus may be invisible at the scalp despite being electrically active. This is why epilepsy surgery planning always requires more than scalp EEG: intracranial recordings (subdural grids, stereo-EEG depth electrodes) are needed to localize the epileptogenic zone for resective surgery. The same principle applies to evoked potential generators: the recorded scalp signal is a far-field approximation that requires anatomic and physiologic knowledge to interpret. The lesson for the clinical electrophysiologist is humility: scalp EEG reveals a smoothed shadow of cortical activity, not its full structure, and over-interpretation of focal scalp findings without clinical correlation is one of the most common sources of false-positive epilepsy diagnoses.

Pitfalls and Pearls

Every recording is a difference: no absolute voltage; the reference matters.
Scalp EEG is spatially smoothed: a 1 cm² source spreads to ~6 cm at the scalp.
Phase reversal in bipolar montage localizes the source between the two channels showing maximum opposite polarity.
Average reference fails for widespread activity: the “reference” becomes contaminated.
Volume conduction: NCS recording electrodes pick up adjacent muscles, not just the target.
Belly-tendon montage: initial positive deflection = misplaced active electrode (not over motor point).
Antidromic sensory NCS is standard for digital sensory studies; larger and cleaner than orthodromic.
Cathode distal: orthodromic motor stimulation; reversed polarity produces partial anodal block.
Equal electrode impedances preserve common-mode rejection; unequal impedances let 60 Hz through.
<5 kΩ impedance for all electrodes; matched within 1–2 kΩ.
Filter settings shape what you see: same data with different filters can look like different pathology.
Notch filter ringing can create artifactual sharp waves around real transients.
Inverse problem: scalp EEG cannot uniquely localize cortical sources; intracranial recording needed for surgical planning.
The 10-20 system uses head proportions, not absolute distances; consistent across patients but each electrode covers a broad cortical area.
Cold limbs slow NCS conduction: always warm to 32°C distally before measuring.
Common artifact patterns: 60 Hz (line), pop (single channel), eye movement (frontal), muscle (frontal/temporal), ECG (vertex/mastoid).

References

Nunez PL, Srinivasan R. Electric Fields of the Brain: The Neurophysics of EEG. 2nd ed. Oxford University Press; 2006.
Ebersole JS, Husain AM, Nordli DR Jr, eds. Current Practice of Clinical Electroencephalography. 4th ed. Wolters Kluwer; 2014.
Preston DC, Shapiro BE. Electromyography and Neuromuscular Disorders. 4th ed. Elsevier; 2021.
Michel CM, Murray MM, Lantz G, et al. EEG source imaging. Clin Neurophysiol. 2004;115(10):2195-2222.
Acharya JN, Hani A, Cheek J, Thirumala P, Tsuchida TN. American Clinical Neurophysiology Society Guideline 2: Guidelines for standard electrode position nomenclature. J Clin Neurophysiol. 2016;33(4):308-311.
Kimura J. Electrodiagnosis in Diseases of Nerve and Muscle. 4th ed. Oxford University Press; 2013.