Functional Neurophysiology

From single neurons to network rhythms

A scalp rhythm is never the product of a single cell or even a single cortical column - it is an emergent property of a network whose nodes are reciprocally connected and whose collective dynamics settle into oscillation. The central insight of functional neurophysiology is that the brain is built from coupled excitatory-inhibitory (E-I) loops at multiple scales, and that oscillation is the natural behavior of such loops. When excitation drives a population, the delayed feedback inhibition it recruits silences that population; the rhythmic interplay of drive and brake sets a frequency. The period of the oscillation is governed chiefly by the kinetics of the inhibition - fast GABA-A-mediated inhibition with a short time constant yields fast (gamma-range) rhythms, while slower inhibitory and intrinsic conductances yield slower bands. Change the strength of the loop, the time constant of the inhibition, the conduction delays, or the level of background excitation, and the frequency shifts. Every band you read is one of these loops operating in a particular regime, and the balance of excitation and inhibition - the E-I balance - is the master parameter that sets where in that regime the network sits.

The thalamocortical-reticular loop as central pacemaker

The thalamus is the metronome of the forebrain. Three populations form the essential circuit: thalamic relay neurons that project to cortex, cortical pyramidal neurons that project back through dense corticothalamic feedback, and the inhibitory thalamic reticular nucleus (TRN), a shell of GABAergic cells that surrounds the thalamus, receives collaterals of both thalamocortical and corticothalamic axons, and gates thalamic output without itself projecting to cortex. The decisive cellular property is the low-threshold, T-type calcium channel expressed by relay and reticular neurons. When these cells are depolarized - as in wakefulness, under cholinergic and aminergic drive - the T-channels are inactivated, the cells fire in tonic single-spike mode, and they faithfully relay sensory information to cortex. When they hyperpolarize - as ascending arousal withdraws during sleep, or under the influence of inhibition - the T-channels de-inactivate, and a small depolarization triggers the low-threshold calcium spike, crowned by a high-frequency burst of sodium action potentials: rhythmic burst-firing mode. It is this burst mode, paced by the reciprocal inhibition between TRN and relay cells and entrained by corticothalamic feedback, that produces the synchronized oscillations of sleep and, in pathology, the spike-wave of absence epilepsy.

This single circuit, operating at different membrane potentials and under different neuromodulatory tone, accounts for a remarkable range of rhythms. The sleep spindle is the TRN-relay loop oscillating in the sigma band during light NREM, with the TRN acting as the pacemaker and corticothalamic feedback synchronizing spindles across wide territories. The alpha rhythm reflects a related thalamocortical resonance in the relaxed waking posterior cortex, with high-order nuclei such as the pulvinar implicated in pacing posterior alpha. And the roughly 3 Hz generalized spike-wave of childhood absence is the same machinery hijacked into pathological hypersynchrony: a corticothalamic oscillation in which cortical hyperexcitability and thalamic burst firing entrain one another across the whole network. The mechanistic unity is clinically actionable - the anti-absence drug ethosuximide works by blocking the very T-type calcium current that underlies the burst mode, while drugs that enhance GABA-B-mediated hyperpolarization can paradoxically worsen absence by deepening the de-inactivation of those same channels. Understanding one loop thus unifies a normal rhythm, a sleep transient, and an epilepsy under a single mechanism, and explains a pharmacological paradox at the bedside.

Why EEG sees what it sees: dipoles, geometry, and synchrony

Two independent variables govern what reaches the scalp, and conflating them is a classic error. Frequency is determined by the timing of the generating loop. Amplitude, by contrast, is determined almost entirely by synchrony and by geometry. The physical source of the scalp signal is the summed extracellular field of postsynaptic potentials in the apical dendrites of cortical pyramidal neurons - not action potentials, which are too brief and too spatially incoherent to sum at the scalp. Because pyramidal neurons are arranged in parallel, perpendicular to the cortical surface, their dendritic currents form open fields that behave like aligned dipoles; when many such dipoles are active in the same direction at the same instant, their fields add. Three consequences follow. First, synchrony sets amplitude: the degree to which many neurons execute the same postsynaptic event at the same moment determines whether their dipoles sum or cancel. Second, geometry matters: a source in a gyral crown projects a clean radial dipole to the overlying scalp, whereas a source buried in a sulcal wall produces a tangential dipole whose maximum appears displaced to the side - the reason a spike can localize away from its generator. Third, scale matters: it is estimated that several square centimeters of cortex must discharge synchronously to produce a deflection visible on the scalp, which is why the scalp EEG is blind to small or deep generators that intracranial electrodes detect easily.

From these biophysics the central paradox of amplitude follows directly. A small population firing vigorously but asynchronously produces a flat scalp trace, while a large population oscillating slowly but in near-perfect lockstep produces high-amplitude waves. This is why deep encephalopathic delta can be larger than healthy alert activity - it is more synchronous, not more metabolically active - and why desynchronization, the replacement of organized rhythm by low-voltage fast activity, is the electrographic signature of cortical activation, whether the physiological arousal that blocks alpha or the focal voltage attenuation that often marks the electrographic onset of a focal seizure. The clinician who equates a high-amplitude record with a healthy brain, or a low-voltage record with a damaged one, has the relationship exactly backwards. Amplitude is a report on synchrony and geometry, filtered by the skull and scalp, not a thermometer of neural activity.

Synchrony, connectivity, and the metrics of the network view

The same principle that sets scalp amplitude scales up into a framework for understanding the whole brain as a system of interacting networks. Functional connectivity describes the statistical coupling between activity at different sites, and the EEG literature uses a graded toolkit to measure it: coherence (spectral correlation between two channels), phase synchronization measures such as the phase-locking value and the (volume-conduction-robust) phase-lag index, and cross-frequency coupling, of which phase-amplitude coupling - the modulation of a fast rhythm's amplitude by the phase of a slower one, as in the coupling of hippocampal gamma to theta - is the most studied. Effective connectivity goes further and attempts to infer directed influence (for example Granger-causal measures), distinguishing which region drives which. A productive way to think about why any of this matters, the communication-through-coherence hypothesis, is that two regions exchange information efficiently only when their excitability cycles are phase-aligned, so that spikes from one arrive at the other during its receptive window. Oscillatory synchrony, on this view, is not an epiphenomenon but a mechanism by which the brain routes information dynamically, binding distributed neural assemblies into transient functional ensembles.

These metrics carry real caveats that the careful reader must hold in mind, and they define the methodological controversies of the field. The dominant pitfall is volume conduction: because a single source is picked up by many electrodes simultaneously, naive coherence and phase-locking estimates can report spurious connectivity that reflects one generator seen from several angles rather than genuine inter-regional coupling - which is why phase-lag-based measures that discard zero-lag interactions are preferred for scalp data. The reference electrode shapes every connectivity estimate, since all signals are differences against a common reference; the choice of montage is therefore not cosmetic but can manufacture or erase apparent coupling. Graph-theoretic summaries (path length, clustering, modularity, hub identification) inherit all of these vulnerabilities and add their own dependence on arbitrary thresholding choices. Source-space analysis after solving the ill-posed inverse problem mitigates volume conduction but introduces leakage of its own. The honest position in 2026 is that EEG connectivity is genuinely informative but methodologically fragile, and that reproducible findings demand volume-conduction-aware metrics, explicit reference handling, and replication - a sobering counterweight to the proliferation of connectivity biomarkers.

The network view of disease in 2026: mechanism and controversy

This network perspective reframes much of clinical electrophysiology, and it is where the field is moving fastest. Epilepsy is increasingly understood as a disorder of pathological hypersynchrony and abnormal network organization rather than a purely local event: the search for the epileptogenic network (rather than a single focus) now informs surgical planning, and intracranial high-frequency oscillations - ripples at 80 to 250 Hz and fast ripples above 250 Hz - are studied as candidate biomarkers of epileptogenic tissue, though their prospective utility for tailoring resection remains an open and actively contested question rather than settled practice. Disorders of consciousness manifest as a breakdown of long-range connectivity and of the brain's capacity for differentiated yet integrated responses; the perturbational complexity index (PCI), computed by perturbing cortex with transcranial magnetic stimulation and measuring the algorithmic (Lempel-Ziv) complexity of the resulting spatiotemporal EEG response, has shown high sensitivity for separating conscious from unconscious states across sleep, anesthesia, and brain injury, and is among the most promising tools for detecting covert consciousness in behaviorally unresponsive patients - while remaining a research instrument whose bedside generalization and thresholds are still being established. Neurodegenerative and psychiatric disease show characteristic, if non-specific, shifts: a slowing of the dominant rhythm and loss of fast activity and connectivity in Alzheimer disease, degradation of sleep spindle-slow oscillation coupling across several dementias, and the gamma-band and E-I-balance abnormalities hypothesized to follow PV-interneuron dysfunction in schizophrenia.

The arc of this pillar is now complete. Normal rhythms are the macroscopic readout of thalamocortical and cortical E-I loops; state determines which regime those loops occupy by setting their neuromodulatory and membrane-potential context; and synchrony, dipole geometry, and connectivity determine what of that activity reaches the scalp and what it means. The waveform on the page is the visible surface of a deep, dynamic, networked system, smoothed and attenuated by the tissues between cortex and electrode. Every expert interpretation is, at bottom, a Bayesian inference about the state of that network from the shadow it casts on the scalp - an inference made under genuine uncertainty, disciplined by mechanism, conditioned on state, and properly humble about the limits of what a few dozen scalp electrodes can resolve.