Brain-machine interface (BMI) is a term used interchangeably with brain-computer interface (BCI) in much of the academic literature. While BCI tends to emphasize computer control (cursor movement, typing, speech synthesis), BMI historically emphasizes direct control of physical machines — robotic arms, powered wheelchairs, exoskeletons, and prosthetic limbs. The distinction is largely one of convention rather than technology.

Origin and Usage

The term BMI gained prominence through the work of Miguel Nicolelis and colleagues at Duke University in the early 2000s. Nicolelis's group demonstrated that ensembles of neurons in the motor cortex of non-human primates could be used to control robotic arms in real time, and they favored the BMI terminology in their publications. Meanwhile, the BrainGate consortium at Brown University, led by John Donoghue and later Leigh Hochberg, used the BCI terminology for similar work.

In practice, the two terms describe the same class of technology: systems that record neural signals, decode intended actions, and translate them into device commands. The IEEE and many journals now treat the terms as synonymous, though individual research groups often have strong preferences.

Current Convention

In the current commercial and regulatory landscape, BCI has become the more widely adopted term. Companies like Neuralink, Synchron, and Precision Neuroscience use BCI in their regulatory filings and marketing materials. The FDA uses "brain-computer interface" in its guidance documents. However, BMI remains common in academic neuroscience publications, particularly those focused on motor prosthetics and robotic control.

Technical Scope

Both BCI and BMI encompass the same core pipeline: signal acquisition (electrodes recording from the brain), signal processing (filtering, feature extraction), decoding (translating neural patterns into commands), and output control (driving a device). Whether that output device is a computer cursor or a robotic arm, the underlying neural engineering principles are identical.