Accurately estimating movement intention from brain waves within 100 milliseconds

The ultimate goal of a brain computer interface (BCI), an interface that reads signals from the brain and connects it to a computer, is to operate machines exactly as the user wishes. Many methods for BCI have been proposed so far, but there has always been the problem of additional sensory stimulation (cognitive load) being necessary, such as long-term training to set up the device in a way that suits the characteristics of each person's brain waves, detection of brain waves generated in response to visual input using images, and so on.

Using the prediction function of the brain, the researchers have conceived a BCI technology to read intended movement (movement intention) from brain waves with high speed and high accuracy, in collaboration with Tokyo Institute of Technology and Osaka University.

Conventional BCI technologies use machine learning algorithms to directly read movement intention from brain waves. These systems can put attentional burden on the user, who has to concentrate on certain thoughts, or certain sensory stimulations in order to enable the system to work. Furthermore, the accuracies in these systems are limited. Here, the conceived BCI technology utilizes the well-known ability of our brain to predict the state of the body after movement are made, or even imagined. The difference between these predictions and the subsequent actual states, called prediction errors, are known to be a crucial determinant of various perceptual and motor abilities in humans, and hence are expected to have a large signature in the brain activity.

Motivated by this hypothesis, the new system proposes to decode what a user intends/imagines by looking for prediction errors between what the user is imagining and a sensory feedback induced using an external stimulator. The results show that this method can be used to decode user intentions with high accuracy and with minimal user training. Since training is not required and the burden is small, it is expected that this method can be applied to an interface for patients with limb paralysis, so that they can operate external devices such as wheelchairs.

Provided by Advanced Industrial Science and Technology