BCI Glossary

International standard for EEG electrode placement. Electrodes are positioned at 10% or 20% intervals between anatomical landmarks (nasion, inion, preauricular points). Locations labeled by brain region (F=frontal, C=central, P=parietal, O=occipital, T=temporal) and hemisphere (odd=left, even=right, z=midline).

Brief (1-2ms) electrical event in which a neuron's membrane potential rapidly rises and falls. The fundamental unit of neural communication. Individual action potentials are not detectable by EEG -- scalp EEG measures the summed postsynaptic potentials of thousands of neurons.

Neural oscillations in the 8-13 Hz frequency band. Dominant during relaxed wakefulness with eyes closed. Alpha power decreases with eye opening, mental effort, or anxiety. Posterior alpha (occipital) relates to visual processing; frontal alpha asymmetry relates to emotional valence.

Process of identifying and removing non-neural signals contaminating EEG data. Common artifacts: eye blinks (EOG), muscle tension (EMG), heartbeat (ECG), electrode movement, and line noise (50/60 Hz). Methods include thresholding, ICA decomposition, and regression-based removal.

Signal processing filter that allows frequencies within a specific range to pass while attenuating frequencies outside that range. In EEG: typically 0.1-100 Hz for raw data, 8-30 Hz for motor imagery, 0.1-30 Hz for ERP analysis. Critical for removing drift and high-frequency noise.

Neural oscillations in the 13-30 Hz frequency band. Associated with active thinking, focus, and motor planning. Excessive beta can indicate anxiety or rumination. Beta desynchronization over motor cortex occurs during movement planning and execution.

Techniques for visualizing brain structure (structural MRI, CT) or function (fMRI, PET, SPECT, EEG, fNIRS, MEG). Functional imaging reveals which brain regions are active during tasks. BCI-relevant modalities prioritize temporal resolution (EEG, MEG) or portability (fNIRS).

A direct communication pathway between the brain and an external device. BCIs acquire brain signals, analyze them in real time, and translate them into commands that control hardware or software. Can be invasive (implanted electrodes) or non-invasive (EEG, fNIRS).

Synonym for BCI, more commonly used in clinical and neuroprosthetic contexts. Emphasizes the machine/prosthetic end of the interface rather than the computational aspect.

Open-source library providing a uniform API for obtaining, parsing, and processing EEG, EMG, ECG, and other biosignal data. Supports 20+ boards including OpenBCI, EMOTIV, Muse, Neurosity, and g.tec. Available in Python, C++, C#, Java, JavaScript, MATLAB, and R.

Number of independent recording electrodes or optodes. More channels provide better spatial resolution but increase setup time, cost, and data volume. Consumer devices: 4-14 channels. Research-grade: 32-256 channels. High-density EEG enables source localization.

A BCI or neurostimulation system that records neural activity, processes it in real time, and delivers stimulation or feedback based on the current brain state. Enables adaptive therapy -- stimulating only when pathological activity is detected. Used in responsive neurostimulation for epilepsy.

Spatial filtering algorithm that maximizes the variance difference between two classes (e.g., left vs. right motor imagery). The most widely used feature extraction method for motor-imagery BCIs. Requires multi-channel EEG (minimum 8-16 channels for effectiveness).

Process of identifying the functional organization of the cerebral cortex. Non-invasive methods use high-density EEG or fNIRS to infer cortical activity. Invasive methods use electrocorticography (ECoG). Used pre-surgically and in BCI development to locate motor, language, and sensory areas.

Invasive neurostimulation technique involving surgically implanted electrodes in specific brain nuclei (subthalamic nucleus, globus pallidus). FDA-approved for Parkinson's disease, essential tremor, OCD, and dystonia. Delivers continuous electrical pulses.

Neural oscillations in the 0.5-4 Hz frequency band. Dominant during deep sleep (Stage 3 NREM). In waking adults, delta activity may indicate brain pathology. Used in sleep staging research and sleep quality assessment.

EEG electrodes that make direct contact with the scalp without conductive gel or saline. Faster setup (seconds vs. minutes) but typically higher impedance and more susceptible to motion artifacts. Materials include metal pins, polymer foams, and carbon nanotubes.

Semi-invasive recording of electrical activity directly from the cortical surface using electrode grids placed under the skull but above the brain. Higher signal quality and bandwidth than EEG. Used in epilepsy monitoring and high-performance BCIs. Also called intracranial EEG (iEEG).

Resistance at the electrode-skin interface, measured in kilohms. Lower impedance means better signal quality. Wet (gel) electrodes achieve < 5 kohm. Dry electrodes typically 10-200 kohm. High impedance increases noise and reduces signal fidelity. Most research systems require impedance below 20 kohm.

Non-invasive method of recording electrical activity of the brain using electrodes placed on the scalp. Measures voltage fluctuations from ionic current within neurons, typically at 0.1-100 Hz. The most widely used modality in consumer and research BCI.

Decrease in power of specific EEG frequency bands time-locked to an event (usually movement or motor imagery). Alpha/mu ERD over sensorimotor cortex indicates cortical activation. The inverse (ERS -- event-related synchronization) indicates cortical idling or inhibition.

Measured brain response directly resulting from a sensory, cognitive, or motor event. Extracted by averaging EEG signals time-locked to repeated stimuli. Components include N100, P200, N200, P300, N400, and P600, each associated with different cognitive processes.

Algorithm that decomposes a time-domain signal into its constituent frequency components. In EEG: computes the power spectrum showing how much energy exists at each frequency band. Foundation of spectral analysis in BCI -- used to extract alpha, beta, theta power features.

Statistical dependencies between neural signals from different brain regions, indicating communication between those regions. Measured using coherence, phase-locking value, or Granger causality. Disrupted connectivity patterns are biomarkers for neurological and psychiatric disorders.

Optical neuroimaging technique that measures brain activity by detecting changes in blood oxygenation using near-infrared light (700-900nm). Slower temporal resolution than EEG but better spatial resolution and immunity to electrical noise.

Neural oscillations above 30 Hz (typically 30-100 Hz). Associated with higher-order cognitive processing, binding of information across brain regions, and consciousness. Gamma power increases during memory encoding, learning, and meditative states.

Blind source separation technique used to decompose multi-channel EEG into statistically independent components. Each component typically corresponds to a distinct neural source or artifact. Standard preprocessing step in EEG research for isolating and removing eye blink and muscle artifacts.

Open-source networking middleware for real-time streaming of time series data, including EEG, with network-level time synchronization. Standard protocol for multi-modal synchronization in research labs. Enables combining EEG with eye tracking, motion capture, and other modalities.

Electrical signal generated by the summed synaptic activity of neurons in a local brain region, recorded by electrodes implanted in brain tissue. Contains information about population-level neural dynamics. Intermediate between single-neuron recordings and EEG.

Non-invasive neuroimaging technique measuring magnetic fields produced by neuronal currents. Requires superconducting quantum interference devices (SQUIDs) in magnetically shielded rooms. Millisecond temporal resolution with better spatial resolution than EEG. Systems cost $2-5M+.

Mental rehearsal of movement without actual execution. Produces detectable changes in sensorimotor rhythm (mu/beta desynchronization). Foundation of motor-imagery BCI systems used in stroke rehabilitation and assistive technology.

EEG oscillation at 8-12 Hz recorded over the sensorimotor cortex (C3/C4 electrode positions). Suppressed (desynchronized) during movement and motor imagery. The primary signal used in motor-imagery BCIs. Named for its arch-shaped waveform morphology.

Computational process of extracting information about intended actions, cognitive states, or perceptual experiences from neural recordings. Uses machine learning (SVMs, CNNs, LSTMs, transformers) to map brain activity patterns to discrete outputs. The core algorithm behind every BCI.

Rhythmic patterns of neural activity across frequency bands (delta, theta, alpha, beta, gamma). Thought to coordinate information processing across brain regions. Oscillation power and coherence are primary features used in BCI classification and neurofeedback.

Real-time self-regulation of brain activity using EEG or fNIRS feedback. The user sees or hears a representation of their brain state and learns to modify it. Used clinically for ADHD, anxiety, and peak performance training.

Alteration of neural activity through targeted delivery of a stimulus (electrical, chemical, or physical) to specific neural sites. Broader than neurostimulation -- includes pharmacological modulation. The therapeutic application of BCI technology for neurological and psychiatric conditions.

The brain's ability to reorganize its structure, function, and connections in response to experience, learning, or injury. The mechanism underlying neurofeedback training, stroke rehabilitation, and BCI skill acquisition. Decreases with age but never reaches zero.

Devices that supplement or replace lost neural function. Examples: cochlear implants (hearing), retinal prostheses (vision), brain-controlled robotic arms. Modern neuroprosthetics use invasive BCIs (Utah arrays, Neuralink) for high-bandwidth motor control.

Umbrella term for techniques that modulate neural activity via electrical, magnetic, optical, or ultrasonic energy. Includes TMS, tDCS, DBS, VNS, and focused ultrasound. Can be invasive or non-invasive. Used therapeutically and experimentally to alter brain function.

Technique using genetically encoded light-sensitive proteins (opsins) to control neural activity with millisecond precision. Requires genetic modification, so currently limited to animal research. Enables cell-type-specific activation or silencing -- far more precise than electrical stimulation.

Event-related potential (ERP) component occurring approximately 300ms after a rare or task-relevant stimulus. Used in BCI spellers (P300 speller), lie detection research, and cognitive assessment. Amplitude reflects attention allocation and working memory updating.

FDA-approved closed-loop neurostimulation system for drug-resistant epilepsy. Continuously monitors brain activity via implanted electrodes and delivers targeted stimulation when pre-seizure patterns are detected. Reduces seizure frequency by 50-70% in clinical trials.

Brain activity recorded when a subject is awake but not performing any specific task. Resting-state EEG/fMRI reveals intrinsic brain network organization (default mode network). Used as a baseline measurement and for biomarker discovery in psychiatric research.

Number of data points recorded per second per channel, measured in Hz. Higher rates capture faster neural dynamics but generate more data. 250 Hz is sufficient for most BCI. 1000+ Hz for high-frequency analysis. Nyquist theorem requires sampling at 2x the frequency of interest.

Compromise between wet and dry electrodes using saline-soaked felt pads or polymer materials that provide moderate conductivity without gel. EMOTIV devices use this approach. Faster setup than wet, better signal than fully dry.

EEG oscillation at 12-15 Hz over the sensorimotor cortex. Associated with motor inhibition and somatosensory processing. SMR neurofeedback (up-training) has been used to treat epilepsy, ADHD, and improve cognitive performance. Overlaps with low beta range.

Mathematical operations applied to raw neural recordings to extract meaningful information. In BCI: filtering (bandpass, notch), artifact rejection, feature extraction, and classification. The signal processing pipeline determines BCI accuracy more than the hardware in most consumer applications.

Signal processing technique for separating the action potential waveforms of individual neurons from multi-electrode recordings. Essential for invasive BCI systems using microelectrode arrays. Modern approaches use template matching, PCA, and machine learning classifiers.

Brain response to visual stimulation at specific frequencies (typically 6-30 Hz). The visual cortex resonates at the stimulus frequency, detectable in EEG. Used in high-speed BCI systems -- SSVEP-based BCIs can achieve 40+ characters per minute.

Neural oscillations in the 4-8 Hz frequency band. Associated with drowsiness, meditation, memory encoding, and hippocampal function. Frontal midline theta increases during focused attention. Excessive theta in children may relate to ADHD.

Neurostimulation technique delivering constant, low-amplitude direct current (1-2 mA) via scalp electrodes. Modulates neuronal excitability -- anodal stimulation increases it, cathodal decreases it. Lower cost and simpler than TMS.

Non-invasive brain stimulation technique using magnetic fields to induce electrical currents in specific brain regions. Used therapeutically for depression (FDA-cleared) and experimentally for cognitive research. Not a recording method -- it stimulates, not reads.

Stimulation of the vagus nerve, typically via an implanted pulse generator on the left vagus nerve in the neck. FDA-approved for epilepsy and treatment-resistant depression. Non-invasive transcutaneous versions (tVNS) are being researched.

Time-frequency analysis method that provides both temporal and spectral information simultaneously, unlike FFT which loses time information. Better suited for non-stationary signals like EEG. Continuous wavelet transform (CWT) shows how frequency content evolves over time.

Traditional EEG electrodes using conductive gel (electrolyte) to reduce impedance at the skin-electrode interface. Gold standard for signal quality. Ag/AgCl (silver/silver chloride) is the most common material. Require trained application and post-session cleanup.