Brain Computer Interface

Brain-Computer Interface Enables Mind-Control Gaming

A study shows a non-invasive one-size-fits-all brain-computer interface solution.

Posted April 3, 2024 | Reviewed by Ray Parker

Key points

Brain-computer interfaces (BCIs) are a new technology that allows users to control computers with their minds.
BCIs work by capturing brain activity and using AI to decode a person's intended actions from that activity.
Researchers have developed a new method for creating BCIs that does not require calibration.

Willi-van-de-Winkel/Pixabay

Innovative technologies are rapidly advancing human capabilities. A new study by researchers at The University of Texas at Austin demonstrates a one-size-fits-all noninvasive brain-computer interface (BCI) that enables users to play computer games through mind-reading.

“Noninvasive brain-computer interfaces (BCI) based on electroencephalography (EEG) have proven efficient in applications such as neurorehabilitation, robotics, communication, or virtual reality,” wrote corresponding authors Satyam Kumar and José del R Millán in collaboration with Hussein Alawieh, Frigyes Samuel Racz, and Rawan Fakhreddine at The University of Texas at Austin.

Using thoughts as a method of managing external computing devices and electronics is a major leap forward for humankind. To put this in context, during the 1950s to 1970s early computers were programmed using stacks of paper punch cards until the rise of magnetic tape, hard drives, and other storage technologies in the early 1980s. In recent times, brain-computer interfaces are gaining traction.

The concept of brain-computer interfaces is simple: use technology to help “read” a person’s mind and execute commands to manage external computing devices. The execution is much more challenging. BCIs require the capturing of brain activity by recordings and figuring out a person’s intended action. The brain activity recording device can be an invasive implant that requires surgery to be placed in the appropriate brain areas or a noninvasive, such as an electroencephalography (EEG).

Artificial intelligence (AI) helps decode a person’s intended activity from their thoughts by predicting intended action in complex brain activity data. A key component to the rise of brain-computer interfaces is the application of AI machine learning to help find patterns in brain activity.

Typically, training a brain-computer interface user involves the time-consuming and painstaking collection of brain activity data to calibrate and construct an individual decoder for each user. As a short cut, decoders use pre-recorded brain data from a single expert, a process known as inter-subject transfer learning.

What sets this study apart is that The University of Texas at Austin researchers have developed an innovative method to produce a brain-computer interface that does not require calibration using inter-subject transfer learning and a special framework that they developed.

The researchers created a Riemannian incremental domain adaptation framework that uses first-order statistics to match the data distributions of the pre-recorded data from a single expert with inexperienced users in real time. In mathematics, Riemannian geometry, or elliptic geometry, is named after German mathematician Bernhard Riemann (1826-1866), whose geometry provided a foundation for Albert Einstein’s general theory of relativity.

For this study, participants wore a noninvasive cap with electrodes that measure electrical brain signals that a decoder translates into commands. The researchers evaluated their framework with 18 healthy study participants on two tasks: a simple game of balancing a digital bar and the Cybathlon car racing game. An initial decoder to translate brain activity into computing commands was developed from data collected from a single user who performed the bar task only, not the car racing game. This decoder worked so well that it enabled the participants to train for both the bar game and the car racing game at the same time, according to the researchers.

“We show that along with improved task oriented BCI performance in both tasks, our frameworks promoted subjects’ ability to acquire individual BCI skills, as the initial neurophysiological control features of an expert subject evolved and became subject specific,” the researchers concluded.

With this innovative framework, it is much easier to generalize brain wave decoders for brain-computer interfaces without time-consuming calibration in the future.