Brain–Computer Interfaces (BCI): Uses and Configurations

A typical BCI contains signal acquisition, preprocessing, feature extraction, decoding, an output device and feedback. The user and the algorithm adapt to one another over time, which makes BCI fundamentally a closed interaction loop between neural activity, signal processing, an external output and feedback. [1,2]

Signals may come from invasive electrodes or from non-invasive modalities such as EEG and fNIRS. For most academic and translational laboratories, EEG remains the most common entry point because it combines high temporal resolution with relatively practical hardware.

Motor imagery means mentally rehearsing a movement without actually performing it. Motor-imagery BCIs decode the associated changes in sensorimotor rhythms—rhythmic EEG activity over sensorimotor cortex that changes during real or imagined movement. P300 is a positive event-related EEG potential that is typically strongest around 300 milliseconds after a task-relevant or infrequent stimulus. P300-based BCIs use this response to infer which item or event the user is attending to. A steady-state visual evoked potential (SSVEP) is a rhythmic EEG response that follows the frequency of a periodically flickering visual stimulus. SSVEP-based BCIs identify which stimulus the user is looking at from the frequency pattern in the EEG. [3]

These paradigms differ in training requirements, visual load, number of possible commands, speed and suitability for different users. The right paradigm depends on whether the goal is communication, device control, feedback or rehabilitation.

BCI systems are designed around an output goal: communication, control, feedback or rehabilitation. The signal and paradigm are selected to match that goal and the abilities of the intended user.

Functional near-infrared spectroscopy (fNIRS) can classify task-related changes in cortical haemodynamics. Its response is slower than EEG because it depends on neurovascular coupling—the process by which local neural activity is followed by changes in blood flow and oxygenation. Hybrid EEG–fNIRS BCIs combine fast electrical and slower haemodynamic information. [4,5]

Hybridization can add complementary features or improve robustness in some settings, but it also increases hardware, synchronization and analysis complexity.

In BCI-based rehabilitation, detected motor intention or motor imagery can be coupled to visual feedback, a robotic device or functional electrical stimulation. Systematic reviews and meta-analyses indicate that such approaches can improve upper-limb motor outcomes in some post-stroke populations, particularly as part of structured rehabilitation programs. [6,7]

The literature nevertheless varies substantially in protocol, training dose, feedback mode and participant selection. The clinical role of BCI rehabilitation should therefore be evaluated for the specific use case and evidence base rather than generalized across populations.

A BCI configuration combines a signal source, paradigm, real-time processing layer, output device and feedback. Non-invasive systems most often use EEG, fNIRS or a hybrid of the two.