The term was coined by Jacques Vidal in 1973, and the idea has stayed the same since then: record a signal from the brain, find the part of it that reflects what the user intends, and act on that part. What has changed is the hardware, the decoding methods, and the range of things a BCI can control. Today that ranges from a spelling interface for someone with locked-in syndrome to a wheelchair, a robotic arm, or a research experiment.
A BCI is not mind reading. It recognises patterns it was trained on, from signals a sensor can actually pick up. That is a narrower thing, and it is why designing the experiment matters as much as the algorithm does.
How a BCI works, step by step
Almost every BCI, invasive or not, runs the same loop. The details change with the signal and the paradigm, but the stages do not.
| Stage | What happens |
|---|---|
| Acquisition | A sensor records brain activity. Scalp EEG measures voltages, fNIRS measures blood oxygenation with light, and implanted electrodes record from the cortex directly. |
| Preprocessing | Filtering, re-referencing and artifact removal strip out line noise, eye blinks and muscle activity that would otherwise dominate the signal. |
| Feature extraction | The cleaned signal is reduced to numbers that carry the intent, such as band power over motor cortex, an evoked response amplitude, or a covariance matrix. |
| Classification | A trained model maps those features to a decision: left hand or right hand, target letter or not, which flicker frequency the user is looking at. |
| Output and feedback | The decision drives a cursor, a speller, a prosthesis or a game, and the user sees the result, which closes the loop. |
Invasive and non-invasive BCIs
Non-invasive
Sensors on or near the scalp: EEG, fNIRS, MEG and fMRI. EEG dominates because it is cheap, portable and fast, at the cost of spatial resolution and signal strength.
Partially invasive
Electrocorticography (ECoG) places electrodes on the surface of the brain under the skull. It gives a much cleaner signal than EEG and needs surgery, usually done for clinical reasons such as epilepsy monitoring.
Invasive
Microelectrode arrays implanted in the cortex record individual neurons. They give the highest-bandwidth control demonstrated so far, and the surgical risk limits them to clinical trials.
Hybrid
Two signals combined, for example EEG with fNIRS, or EEG with EMG. Each one covers a weakness of the other. See multimodal BCI.
The three classic paradigms
A paradigm is the mental task or stimulus that produces a decodable signal. Most non-invasive BCI research uses one of three.
| Paradigm | What the user does | What is decoded |
|---|---|---|
| Motor imagery | Imagines moving a hand, the feet or the tongue | Changes in mu and beta rhythm power over the motor cortex |
| P300 | Attends to a rare target among frequent flashes | A positive brain response about 300 ms after the target |
| SSVEP | Looks at one of several lights flickering at different rates | Brain activity at the flicker frequency over visual cortex |
What BCIs are used for
Communication
Spellers and selection interfaces for people with ALS, locked-in syndrome or severe paralysis. This is the original clinical motivation.
Rehabilitation
Motor imagery BCIs paired with feedback or stimulation in stroke rehabilitation, where the brain signal triggers the movement assistance.
Control
Wheelchairs, robotic arms, prostheses and smart-home devices driven by decoded intent.
Research
Decoding is also a measurement tool: if a classifier can tell two conditions apart from brain activity, the brain is encoding the difference.
Passive monitoring
Estimating workload, fatigue or attention without the user issuing a command at all.
Gaming and consumer
Low-channel headsets for focus training and interaction. These have a far smaller signal budget than lab systems.
Building one yourself
Most of the work in a BCI study happens offline: recording sessions, cleaning the data, extracting features, training a decoder, and finding out honestly how well it generalises to a new session or a new person. That last step is where published accuracies most often fall apart.
BCILattice is a desktop application for that offline half. It imports EEG, fNIRS and EMG recordings, runs preprocessing and analysis, and trains classical and deep decoders on your own machine with evaluation guardrails built in. To learn the field from the ground up, start with BCI for beginners.
Questions
What is a brain-computer interface in simple terms?
A brain-computer interface is a system that reads brain activity with a sensor, recognises patterns that reflect what the user intends, and turns them into a command for a computer or device, without the user moving.
Can a BCI read my thoughts?
No. A BCI recognises a small number of patterns it was trained on, such as imagined left versus right hand movement, from signals a sensor can pick up. It cannot decode arbitrary thoughts.
What is the difference between invasive and non-invasive BCIs?
Non-invasive BCIs measure from outside the skull, usually with EEG, and need no surgery. Invasive BCIs implant electrodes in or on the brain, which gives a much stronger, more detailed signal but carries surgical risk.
How accurate are EEG-based BCIs?
It depends on the paradigm, the number of classes, the user and the evaluation. Two-class motor imagery and P300 spellers often work well for many users, while some users never achieve reliable motor imagery control. Reported accuracies should always be read alongside how the data was split for testing.
Do I need to know programming to build a BCI?
Not for the offline analysis and decoding work. Graphical tools like BCILattice cover preprocessing, feature extraction and model training without code. Real-time control systems usually need some engineering.
Which signal is best for a BCI?
EEG is the most common because it is affordable, portable and has millisecond timing. fNIRS is slower but more tolerant of movement. Implanted electrodes give the best signal but require surgery.
Read the reference
Related
How EEG Works: What an EEG Measures
Electroencephalography (EEG) records the brain's electrical activity with electrodes on the scalp. It has millisecond timing and centimetre-scale blur, and knowing both is the key to using it well.
BCI for Beginners: How to Get Started with EEG and BCI
You do not need a headset to start. A public dataset, a laptop and a clear first question will teach you more than a new device and no plan.
Motor Imagery BCI: EEG Classification Explained
Imagining a movement suppresses the mu and beta rhythms over the opposite motor cortex. A motor imagery BCI detects that suppression and turns it into a command, with no movement and no external stimulus.
BCI Software and Tools Compared
Some BCI tools run a live system, some are libraries you program against, and some carry an offline study from recording to result. Most labs need more than one.