A brain-computer interface connects activity in the nervous system to a computer or another device. Sensors record signals associated with attempted movement, speech or another task. Software then translates patterns in those signals into commands, such as moving a pointer, selecting a letter or producing speech.
BCIs can be non-invasive, with sensors placed on the scalp, or invasive, using electrodes positioned on or in the brain. Each approach involves a trade-off between signal detail, surgical risk, convenience and long-term stability. Some systems require repeated calibration because recordings can change over time.
The strongest clinical research has often focused on restoring communication or control for people with paralysis or severe speech impairment. Demonstrations may show impressive results for one participant under closely supported conditions, but that does not prove a device will work equally well for everyone or during independent everyday use.
When comparing BCIs, look for the number of participants, recording method, measured accuracy and speed, hours of unsupervised use, adverse events and trial stage. A successful research demonstration is not the same as regulatory approval or routine medical availability.