How Brain-Machine Interfaces Engage Neural Plasticity


by Kayt Sukel

February 15, 2017

Over the past year, scientists have made great strides in the development of brain-machine interfaces (BMIs), wired external devices that are controlled solely by brain activity [see Roadmapping the Adoption of Brain-Machine Interfaces”]. Last October, Nathan Copeland, a man who had been paralyzed from the chest down for more than 10 years, made headlines when he fist-bumped President Obama with a BMI-controlled robotic arm using only his thoughts. As BMI-related technologies and neuroprosthetics become more sophisticated, researchers are learning that these tools can make some fascinating changes to the brain, engaging its natural plasticity in sometimes unanticipated ways. Understanding those changes to underlying plasticity, some say, could offer clues to how to rewire and rehabilitate the damaged brain—perhaps even without the need of external hardware.

From motor to visual

Prosthetics, even without the addition of a BMI component, can alter the brain’s connections, says Lewis Wheaton, director of the Cognitive Motor Control Lab at the Georgia Institute of Technology says.

“Use of a prosthetic device makes a kind of switch in brain function without any kind of training at all, in particular with motor planning and motor execution. We see a similar change in activation pattern not just in amputees wearing a prosthetic device but individuals with sound limbs using a prosthetic device as well,” he says. “Typically, this shift involves motor planning being taken out of what we generally think of as the canonical motor circuit in motor cortex and that activity moving into more of the visuo-spatial processing areas. This remapping happens fairly quickly.”

Wheaton says this pattern of activity is not unlike what you see when you look at the brains of people who are learning to use novel tools for the first time. “When we exposed younger undergrads to tools that they’d never used before, old tools that people used many years ago, and then looked at their brain activity when they tried to use them, we saw them utilizing that same visuo-spatial network,” he says. “Eventually, with lots of practice, the use of those tools can get encoded into the canonical motor circuit. But that doesn’t happen with prosthetics. And why that doesn’t happen with a prosthetic is interesting. It may be because prosthetics aren’t giving people the right kinds of sensory feedback. And it opens the broader question of how important sensory feedback is to anchoring the use of prosthetics and BMIs into the motor circuitry.”

Wheaton is studying how giving some haptic sensation can help with prosthetic adaptation. If we can better understand how the brain is translating this sensory information, we may not only be able to design better, smarter BMIs and prosthetics in the future but also design more effective rehabilitative techniques for people who have had damage to the brain itself.

“Our goal is to improve function,” he says. “And these changes give us a window into a system that is injured generally in the periphery. That is, the brain is typically intact in amputees. But if we can understand the changes and neuroplasticity in a system that is intact, then we have a better chance of understanding what might be going wrong, and how we can best intervene, when the central system is affected like in stroke or brain injury. We can find out the right types of sensation that can help restore function, or perhaps even harness the brain’s dependency on visual feedback that we see in some way to shift the neural activity back to motor cortex.”

Secrets of the decoder

Jennifer Collinger, a member of the University of Pittsburgh Medical Center (UPMC) team who helped design the system that gave Obama his fist-bump last year, also thinks that sensation can help with prosthetic acceptance. She and her team electrically stimulated the cervical dorsal root ganglia (the cluster of sensory nerves that carry information from the peripheral nervous system to the brain) in different cervical bones in the spine in upper limb amputees. The amputees reported that this stimulation could evoke feelings that seemed to come from their missing limbs. The UPMC team also has been able to create feelings of sensation in its BMI study participants—when the robotic arm is touched, the person connected to the robotic arm through a neural decoder can register that touch in the brain.

Collinger says that the group also has seen some changes in brain activity after the use of these different stimulators or BMI systems. In fact, with the robotic arm BMI, they often have to retrain that decoder, the part of the system that translates neural activity into commands for the robotic arm, after about six hours.

“There are a lot of different factors that may lead to us having to recalibrate our systems—the arrays may move and we’re not recording from the same neurons” for one, she says. “But when learning occurs, you can also expect to see changes in activity then. And we have seen that in some of our participants. We don’t necessarily have a good handle on just how flexible the neural activity is, or why it may be changing.”

Jose Contreras-Vidal, a researcher at the University of Houston, says that mounting evidence is showing that BMI systems may promote plasticity in the brain, which could ultimately help scientists better inform rehabilitative medicine. He and his colleagues had study participants use an electroencephalogram (EEG) type BMI to control a computer avatar as it walked in a virtual environment. They discovered that its use resulted in cortical adaptations: With practice, the participants gained more control of the avatar using their own brain signals.

“BMI systems can promote plasticity in the brain as well as change different physiological systems in the body. This opens the possibility that we could use BMI to retrain movement and to optimize the design of BMI systems in the future to retrain specific cortical areas,” he says. “This could help us design more effective interventions for rehabbing cognitive-motor function in patients in the future, too.”

Early days, high hopes

Wheaton says that while BMIs and other neuroprosthetics are helping us to better understand the plasticity involved in training motor procedures, he cautions that this research is still in its early days and we have much yet to learn. Any type of neuromodulator, whether a BMI, a deep brain stimulator, or pain neuromodulating device, may change the brain’s physiology and functional anatomy in similar or different ways.

 “All of these studies give us some really nice concepts to consider. Can we use BMI-related approaches to enervate nerves? What would that coding look like? How would it function? How might it change with learning? If we patched into a set of nerves, would it change over time as you get used to whatever you’re doing with the device?” he says. “If there is a change, is that always good? Is it always bad? When is it good or bad? There’s a lot of great data out there to consider. But there are still a lot of questions we need to answer, too.”

Collinger agrees. “If we can better understand how BMIs or other prosthetics can change the brain, it’s possible that we could directly repair the injured spinal cord or the damaged pathways in stroke. Certainly, different people are working on those potential medical treatments now,” she says. “If we can one day develop medical therapies that can regrow axons and fix these damaged motor connections, a better understanding of how the brain changes in response to BMIs or other rehabilitative techniques could help us figure out how to grow back those connections in a way that makes functional sense and can actually help restore movement. That’s the holy grail, really.”