Electroencephalography (EEG) is the oldest and most widely available method for recording brain electrical activity. Electrodes placed on the scalp detect voltage fluctuations generated by the synchronous activity of large populations of cortical neurons. EEG is the foundation of non-invasive BCI systems and remains the standard of care for clinical applications including epilepsy diagnosis, sleep staging, and intraoperative monitoring.

How EEG Works

Scalp EEG electrodes detect the summated postsynaptic potentials of tens of thousands of cortical pyramidal neurons oriented perpendicular to the cortical surface. The signal passes through cerebrospinal fluid, skull bone, and scalp tissue before reaching the electrode — each layer attenuating and spatially smearing the signal. As a result, EEG has relatively poor spatial resolution (approximately 1-2 cm) compared to invasive methods but offers millisecond-level temporal resolution.

EEG for BCI

EEG-based BCIs exploit several neural phenomena:

  • P300 speller: The user focuses on a target letter in a flashing grid. When the target row or column flashes, a P300 event-related potential is evoked. The BCI detects which flash produced the P300 and selects the corresponding letter. Typical speed: 2-8 characters per minute.
  • SSVEP: The user gazes at a flickering stimulus, and the visual cortex produces oscillations at the flicker frequency. Multiple stimuli at different frequencies enable multi-choice selection. Faster than P300 spellers.
  • Motor imagery: The user imagines moving a hand, foot, or tongue, producing lateralized changes in sensorimotor rhythms (mu/beta desynchronization). These patterns are classified to control a cursor or wheelchair.
  • Error-related potentials (ErrP): The brain produces a characteristic EEG signature when it detects an error, which can be used to automatically correct BCI mistakes.

Limitations for BCI

EEG is limited by low signal-to-noise ratio, susceptibility to muscle artifact, low spatial resolution, and the need for conductive gel or saline (wet electrodes) in most systems. Information transfer rates for EEG BCIs are typically 10-50 bits per minute — orders of magnitude lower than intracortical BCIs. However, EEG requires no surgery, is inexpensive, and is available today for consumer and clinical use.

Clinical and Consumer Landscape

Clinical EEG systems are manufactured by Natus (Neuroworks), Compumedics, and others. Consumer-grade EEG devices for BCI and neurofeedback include Emotiv EPOC X, OpenBCI, Muse, and Cognixion ONE. The gap between consumer and clinical EEG in terms of signal quality, channel count, and reliability remains significant.