Neurophysiological Signals

Neurophysiological signals are observable and recordable manifestations associated with nervous-system activity. These include electrical potentials, magnetic fields, extracellular field activity, action-potential-related events, and neurovascular responses. Such signals provide temporally and spatially structured evidence during behavior but are not the nervous system itself. A neural process is a biological event or activity occurring within the nervous system. A neurophysiological signal is a measurable manifestation of one or more neural processes captured through biophysical and instrumental means. A recorded channel is a single measurement trace obtained from a sensor or electrode. A derived neural measure is a representation or feature computed from raw recordings, such as spectral power or source estimates. A behavioral interpretation is a scientifically defined hypothesis or claim relating neural measures to observable behavior or psychological constructs. Importantly, a neurophysiological signal is not a direct measurement of perception, attention, memory, emotion, intention, consciousness, diagnosis, or any other behavioral or psychological construct.

Meaning of Neurophysiological Signals

Neurophysiological signals are measured consequences of neural activity observed through specific biophysical mechanisms. Electrical recordings measure potential differences or extracellular voltages generated by transmembrane and return currents. Magnetic recordings measure fields generated by neural current flow. Neurovascular recordings observe vascular and hemoglobin changes coupled indirectly to neural activity. These signal families differ in physical origin, temporal resolution, spatial sensitivity, and inferential distance from neuronal events.

A neural event such as synaptic activity, neuronal firing, or membrane current can contribute to several observable signals. Conversely, one recorded neurophysiological signal can contain mixtures of many neural and non-neural sources. Terms must be carefully distinguished:

• Neural activity: Biological processes occurring within neurons or networks (e.g., firing, synaptic currents).
• Neural signal: The underlying biophysical manifestation of neural activity (e.g., electric currents, magnetic fields).
• Neurophysiological signal: The measurable output captured by sensors reflecting neural signals plus noise and confounds.
• Neuronal firing: Action potentials or spikes generated by neurons.
• Synaptic activity: Transmembrane currents at synapses related to neurotransmitter action.
• Field potential: Summed extracellular potentials generated by local populations of neurons.
• Sensor-level recording: Electrical, magnetic, or optical data recorded at a sensor or electrode location.
• Source estimate: An inferred neural generator derived from sensor data using computational models.
• Neural measure: A derived quantity characterizing aspects of neural signals (e.g., power spectrum).
• Behavioral construct: A defined psychological or behavioral phenomenon (e.g., attention, motor preparation).

The same neural event can manifest in multiple signals, and no single signal represents the entire nervous system or a psychological construct directly.

Neural refers specifically to nervous-system processes. Neurophysiological evidence concerns measurable manifestations of those processes. Broader physiological evidence may include signals regulated by neural activity but not direct neural measurements (e.g., peripheral autonomic responses, muscle activity). Peripheral autonomic or muscular signals should not be relabeled as direct neural recordings simply because they are influenced by the nervous system.

Historical Foundations of Neurophysiological Recording

In 1875, Richard Caton reported electrical activity recorded from the exposed cerebral cortex of animals, demonstrating early that brain activity could produce measurable electrical variation. This work laid foundational evidence for electrophysiological brain recording, although the currents observed were not identical to modern scalp EEG, nor did this single experiment create the entire discipline.

Hans Berger pioneered human electroencephalographic work in the 1920s, publishing the first systematic report of the human electroencephalogram in 1929. His observations of alpha and beta rhythms helped establish that electrical activity recorded noninvasively from the human head contained reproducible physiological structure. However, these rhythms were not direct representations of thoughts or mental content.

In 1934, Edgar Adrian and Brian Matthews independently reproduced and validated Berger's alpha-rhythm findings. This episode marked a scientific transition from a controversial trace that could be dismissed as artifact toward reproducible human neurophysiological evidence sensitive to behavioral conditions such as eye opening and visual engagement.

David Cohen's 1968 recording of the magnetic field generated by human brain activity marked the beginning of magnetoencephalography (MEG). Subsequent improvements in sensitivity with SQUID-based detectors enhanced MEG’s utility. This development showed that the same underlying neural currents can produce both electric potentials and magnetic fields, while EEG and MEG measurements remain physically and spatially distinct.

Biophysical Origins of Neurophysiological Signals

Neuronal membrane potential reflects the voltage difference across the neuronal membrane. Transmembrane ionic currents flow through ion channels and pumps, generating electrical activity. Synaptic currents are transmembrane currents at synapses triggered by neurotransmitter release. Action potentials are brief regenerative membrane events associated with neuronal firing, characterized by rapid depolarization and repolarization.

Macroscopic EEG and MEG signals are dominated by the spatial and temporal summation of postsynaptic and dendritic currents across neuronal populations rather than by isolated axonal spikes.

Extracellular fields arise because neural current sources generate electrical potentials that spread through brain tissue, cerebrospinal fluid, skull, scalp, and other conductive structures—a process called volume conduction. This spatial mixing means that one neural source can influence multiple sensors, and one sensor can contain contributions from multiple sources.

Neuromagnetic fields are generated by neural current flow and their detectability depends on source geometry, orientation, synchrony, sensor sensitivity, distance, and environmental shielding. EEG and MEG can respond differently to the same underlying cortical source geometry and should not be treated as interchangeable recordings.

Neurovascular coupling is a different evidential pathway: local neural activity can alter metabolic demand, cerebral blood flow, blood volume, and oxygenated and deoxygenated hemoglobin concentrations. Measurements based on these vascular consequences are temporally slower and more indirect than electrophysiological recordings. A hemodynamic response should never be equated directly with neuronal firing.

Electrical and Magnetic Brain Signals

Scalp EEG records voltage differences at the head surface using electrodes referenced to a designated site or montage. It primarily reflects spatially summed neural current sources after substantial volume conduction and filtering by head tissues. Electrode location indicates sensor location but does not guarantee that activity originated directly underneath that electrode.

MEG measures magnetic fields produced by neural current flow using highly sensitive magnetometers. It has millisecond-scale temporal sensitivity similar to EEG but differs in sensitivity to source orientation and interactions with head conductivity. MEG does not provide uniquely exact or direct source localization.

Intracranial electrophysiological recording broadly refers to electrical measurements using electrodes placed within or directly on neural tissue. Electrocorticography (ECoG) involves electrodes on the cortical surface, while depth-electrode recordings sample deeper brain structures. Proximity to neural sources can improve spatial selectivity and signal amplitude but remains highly dependent on electrode geometry and sampling location. Intracranial recordings do not observe the entire brain.

Local field potentials (LFPs) emphasize lower-frequency extracellular population activity strongly shaped by synaptic and other transmembrane currents. Spike recordings emphasize brief action-potential-related events from nearby neurons. Although related, LFPs and spikes describe different aspects and scales of neural activity.

Electrical recordings measure voltage differences relative to a reference. Changing the reference or montage can alter channel waveforms, amplitudes, polarity, and apparent spatial relationships without changing underlying neural sources. The choice of reference is thus an integral part of the measurement definition.

Neurovascular and Optical Brain Signals

Functional near-infrared spectroscopy (fNIRS) is an optical method estimating changes in oxygenated and deoxygenated hemoglobin concentrations in superficial cortical tissue by measuring near-infrared light absorption. fNIRS relies on neurovascular coupling and thus indirectly measures neural activity with slower response dynamics than EEG or MEG.

Oxyhemoglobin change, deoxyhemoglobin change, total hemoglobin-related change, cerebral blood flow, and neural activity are physiologically related but distinct variables. Extracerebral blood flow, systemic physiology, optode coupling, motion artifacts, and scalp hemodynamics can influence fNIRS recordings and must not be conflated with neural activity.

Electrophysiological signals (EEG, MEG) follow neural activity on millisecond timescales, whereas the neurovascular response develops over seconds. Comparing peak times across these signal families as if they represent the same physiological event is inappropriate.

Ongoing, Evoked, and Event-Related Activity

Ongoing neural activity is spontaneous or background neural signal present independently of defined stimuli or events. Activity temporally associated with specific stimuli, actions, decisions, or feedback events occurs alongside ongoing activity and does not imply neural silence before the event.

Evoked or event-related potentials (ERPs) and fields (ERFs) are reproducible changes estimated by temporally aligning signals to defined events and averaging or modeling across multiple observations. An ERP or ERF is a derived estimate representing event-related structure rather than the literal waveform of every individual trial.

Here, ( x_k(t) ) is the time-aligned signal for observation ( k ), and ( N ) is the number of observations. Averaging emphasizes components consistently time-locked across observations while attenuating non-time-locked variation. This average does not imply that the same component occurred with identical latency or amplitude in every observation.

Latency, polarity, amplitude, duration, topography, and component terminology describe properties of event-related responses. Component names based on polarity and nominal latency are conventions, not direct identities with single cognitive processes. Observed latency depends on event definition, filtering, referencing, averaging, and physiological variability.

Oscillatory and Spectral Neurophysiology

Neural oscillatory activity refers to rhythmic or approximately rhythmic population-level variation observable in electrical or magnetic recordings. Representations of oscillatory structure include spectral power, dominant frequency, phase, phase consistency, and time-varying spectral content. Frequency-band labels are analytical conventions, not single neural mechanisms.

Historically established examples, such as the alpha rhythm, illustrate that oscillatory activity can change with behavioral condition. Alpha activity is not homogeneous across all brain regions, people, tasks, or recording conditions. Changes in band power should not be directly mapped to attention, inhibition, relaxation, engagement, or other constructs without context.

Event-related synchronization (ERS) and desynchronization (ERD) describe increases or decreases in oscillatory power relative to a defined reference condition or interval. These terms describe measured rhythmic activity changes but do not identify the neural mechanism or behavioral function causing the change.

Phase describes position within an oscillatory cycle under an analytical representation. Phase consistency or phase locking describes reproducibility of phase relationships across time, events, or signals. Phase locking should not be equated by definition with communication, connectivity, causality, or information transfer.

Spatial Representation and Source Inference

Sensor space consists of signals measured directly at electrodes, magnetometers, optodes, or intracranial contacts. Source space representations estimate underlying neural generators using forward models, inverse methods, anatomical assumptions, regularization, or related constraints. A source estimate is an inference, not a directly observed brain current.

The inverse problem arises because multiple different neural source configurations can generate similar sensor-level electrical or magnetic patterns. Inferring unique neural generators requires additional assumptions. Increasing sensor count or prettier spatial maps does not remove this fundamental nonuniqueness.

Topography refers to scalp maps or sensor distributions visualizing the spatial pattern of measured signals at sensors. These are not direct anatomical maps of neural activation. Volume conduction, reference choice, source orientation, head anatomy, interpolation, and sensor geometry all shape observed patterns.

Neural Relationships and Connectivity

Anatomical connectivity concerns physical neural pathways linking brain regions. Functional connectivity refers to statistical dependence among signals or estimated sources. Effective or directed connectivity attempts to characterize directed influence under additional assumptions.

Measures such as correlation, coherence, or phase locking are not synonyms for anatomical connection or causation.

Volume conduction and field-spread confounds affect connectivity analysis because a common neural source can appear simultaneously at several sensors, creating apparent correlation, coherence, or phase relationships without direct interaction between underlying neural populations. This is especially significant for scalp-level electrophysiological connectivity.

Common-input and unobserved-source problems occur when two neural signals covary because both receive influence from another process. Directed relationships can be distorted by hidden variables, measurement mixing, unequal delays, and preprocessing. Neural connectivity estimates should be treated as model-dependent evidence rather than proof of communication or causal mechanism.

Measurement Conditions and Non-Neural Contributions

Neurophysiological recordings contain non-neural and artifactual contributions. These include eye movements and blinks, facial and scalp muscle activity, cardiac activity, respiration, movement, electrode or optode motion, cable motion, environmental electromagnetic interference, poor contact, and instrumentation artifacts. Physiological signals from the body are not inherently "noise"; their status depends on the neural measurement objective.

Biological and contextual determinants influencing recorded neurophysiology include vigilance, sleep, posture, movement, medication or substances, age-related factors, sensory environment, task structure, and participant-specific anatomy. These influences are not diagnostic categories but important factors to consider.

Preprocessing choices such as referencing, filtering, artifact correction, epoching, baseline treatment, channel rejection, source reconstruction, and decomposition can materially alter waveform, spectrum, timing, topography, and connectivity estimates. No preprocessing choice is universally neutral or correct; all alter interpretive outcomes.

Neurophysiological Signals and Behavioral Meaning

Neurophysiological signals can be behaviorally informative because nervous-system activity participates in perception, action preparation, motor control, learning, memory, attention, decision processes, affective responses, sleep, social interaction, and many other behaviors. Recorded neurophysiology can therefore provide temporally and spatially structured evidence about neural processes associated with behavior when evidential relationships are explicitly defined.

Interpretation involves many-to-many mappings: one behavioral condition can involve multiple neural processes and signal patterns, while similar neurophysiological changes can appear under diverse behavioral conditions. Universal mappings from one neural feature or spatial location to one mental state are scientifically unjustified.

Distinctions include:

• Neural correlate: A statistical association between a neural measure and behavior.
• Neural marker: A feature predictive of behavior without implying causality.
• Predictor: A variable used to estimate behavioral outcomes.
• Mechanism: A causal process explaining behavior.
• Explanation: A scientific account linking neural activity to behavior.

Statistical association alone does not establish that a neural feature is necessary, sufficient, causal, mechanistically explanatory, or uniquely specific to the behavior.

Use in Behavioral Signal Processing

Neurophysiological signals are useful in Behavioral Signal Processing because they provide millisecond-scale evidence of neural dynamics via electrophysiology and magnetophysiology, spatially informative invasive recordings where available, and slower neurovascular evidence through optical methods. Their value lies in relating nervous-system dynamics to behavior without assuming that neural measurement provides a privileged or complete representation of mind.

Representative uses include:

• Perception and attention research: Temporal dynamics of sensory processing and attentional modulation.
• Action preparation and motor behavior: Neural signatures preceding movement initiation.
• Learning and memory: Changes in cortical oscillations and connectivity during acquisition.
• Workload-related research: Neural indices reflecting cognitive load or fatigue.
• Affect-related behavior: Neural correlates of emotional responses.
• Sleep and vigilance: Oscillatory patterns characterizing sleep stages and alertness.
• Language-related processing: Event-related potentials elicited by linguistic stimuli.
• Social interaction: Neural dynamics during joint attention or communication.
• Human-computer interaction and neuroadaptive systems: Real-time neural feedback for interface adaptation.
• Brain-computer interfaces (BCIs): Translation of neural signals into control commands.
• Rehabilitation-related research: Neural markers tracking recovery or therapy effects.
• Naturalistic behavioral observation: Neural signatures during real-world activities.

Neurophysiological quantities can serve as predictors, outcomes, reference evidence, contextual evidence, control signals, or behavioral correlates depending on the scientific question. An ERP amplitude, oscillatory measure, source estimate, spike rate, connectivity estimate, or hemodynamic response can play various analytical roles, which must be explicitly defined.

Relationships with cardiovascular activity, respiration, electrodermal activity, electromyographic activity, body movement, gaze, facial behavior, vocal behavior, language, task events, and environmental conditions may inform interpretation but do not provide automatic validation or disconfirmation of neurophysiological evidence.

Quantification and Scientific Limits

Neurophysiological analysis employs waveform analysis, event-related averaging, spectral and time-frequency methods, spike and firing-rate analysis, source estimation, spatial filtering, connectivity analysis, nonlinear dynamics, statistics, probability, and machine learning. The event-related averaging equation introduced earlier describes one estimation operation and must not be generalized as an equation of cognition or behavioral meaning.

Inferential distance refers to the gap between measured neurophysiological evidence and behavioral or psychological claims. Claims about measured voltage, magnetic field, spike timing, local field potential, spectral power, ERP amplitude, source estimate, connectivity statistic, or hemoglobin change are closer to neurophysiological evidence than claims about attention, memory, emotion, workload, intention, awareness, decision, deception, personality, diagnosis, or subjective experience. Stronger behavioral claims require explicit operationalization, context, suitable reference evidence, validation, and consideration of alternative neural and non-neural explanations.

Computational neurophysiological analysis carries risks of unintended information and confounding. Models can exploit participant identity, anatomy, recording site, reference configuration, sensor quality, eye or muscle artifact, movement, task timing, stimulus structure, preprocessing choices, session, vigilance, missingness, or equipment-specific patterns while appearing to predict a behavioral target. Predictive performance does not establish that the intended neural pathway or behavioral mechanism has been identified.

Larger neural responses, stronger spectral power, greater connectivity, more focal source estimates, higher firing rates, or larger hemodynamic changes have no universal behavioral valence. Interpretation must consider the recording principle, neural process, anatomical context, individual, task, baseline, timescale, analysis method, and scientific question.

In synthesis, neurophysiological signals are electrical, magnetic, extracellular, spiking, and neurovascular manifestations associated with nervous-system activity and observed through specific biophysical and instrumental relationships. Scientific interpretation requires separating neural process, physical manifestation, sensor-level recording, derived representation, estimated source or relationship, behavioral cue, and behavioral claim. This evidential chain preserves rigor and avoids treating recorded brain signals as direct access to mental content.