Camille Jeunet-Kelway: How neurofeedback research can help advance brain-computer interfaces

The idea that we might, someday, control things through a chip that reads our brain is grabbing attention. Brain-computer interfaces, or BCIs as they are known, are not science fiction - they really work and have the potential to help people control assistive devices. At CYBATHLON BCIs have their own discipline, an event where pilots must solve tasks in a digital scenario by sending the appropriate commands at the right time using only their brain. Camille Jeunet-Kelway, a CNRS Research Scientist, tells us more about the trends to watch in BCIs.

Welcome Camille! Please can you give us a brief introduction to you, your professional background, and your current research?

I did a degree in Applied Mathematics and then a Masters and PhD in Cognitive Sciences. Since 2018, I have a tenured position working in the Institute of Cognitive Neuroscience in Bordeaux. My main research focus is using BCIs and neurofeedback training procedures to improve and restore cognitive and motor skills.

Can you describe what a brain-computer interface is, and how you use it in your research?

BCIs are technologies that enable the recording, processing and translation of brain activity. Active BCIs work on a closed loop where brain activity is recorded then fed back to users in real time. This neurofeedback can be used to control applications or self-regulate specific brain activities, such as sending mental commands to assistive technologies or helping users restore or improve cognitive or motor abilities. In my work I use neurofeedback training for this second aspect, improving or restoring motor skills.

Your research involves athletes and people who have suffered from strokes, can you explain a bit more about how BCI technologies can help to improve motor skills in these patients?

All my research using BCIs relies on a concept called “neurofunctional equivalents”. This is the fact that there are similar brain networks involved when you perform a movement or when you imagine this movement - just thinking about movements will solicit the motor cortex.

In stroke patients we use this to try and restore connections or reinforce new ones in the motor cortex, so that they can begin to move again. The same principle can be applied for athletes who want to improve their motor skills or self confidence.

Your application of BCI technology for motor skill rehabilitation is very different to the digital game applications in CYBATHLON competitions. How do you see the use of BCIs developing over time?

Applications such as the control of assistive technologies are really at the core of BCI research. In CYBATHLON, there are many different disciplines, one of them is the control of digital applications through BCIs, for example sending mental commands to video games.

I would say that we are not there yet for using these applications in everyday life because they are not reliable enough and the training to get good performances is quite long.

But the good news is that we are improving the sensors, the machine learning algorithms, our understanding and the training protocols so that in the coming years we could see those performances raise a lot. There is also a lot of public and private investment and more awareness through media coverage, especially around events like CYBATHLON.

Where do you see the challenges in improving the efficiency and reliability of BCIs, both as a neurofeedback tool, and in the world of assistive technology?

BCIs require the brain and a computer, so there are challenges on the machine side and on the human side. On the machine side, we really need to improve the quality of the sensors to capture high quality brain activity.

Then we need to improve the translation of the signals into commands and into feedback using machine learning for signal processing. For controlling applications and assistive technology we really need close to 100% accuracy because if you send a command to go to the left or go to the right to your wheelchair you want it to be correct.

It is less of a problem in neurofeedback training if you don't always capture good information. Because of that I would say that we're closer to everyday life use of neurofeedback procedures for stroke rehabilitation or attention disorders, where there is strong evidence that BCIs are useful.

However, even if you have the best sensors and the best AI algorithms, a BCI will not work if the person is not able to produce clear signals. To ensure this, we need to improve our understanding of the neurophysiological and the cognitive aspects of human performance and learning in BCIs.

There is also critical work that needs to be done on the usability and acceptance of BCIs by the patient, caregiver and general population, and we need to really think deeply about the ethics and regulations around their use.

How do you think events like CYBATHLON help to promote research and development in key areas, such as BCI technology?

CYBATHLON shows that BCIs are not science fiction, that they can work, but it also shows the reality that there are still challenges to overcome. I think it is very valuable for the research community and potential users to have a precise idea of where we are at the moment.

CYBATHLON also provides great visibility of the field to the general public, the industry and clinicians. People gather together from different disciplines that may not otherwise interact at all. That is very motivating and can potentially attract new talents to the field.

You can see the potential. In a few years, I would not be surprised to see these technologies in everyday use because they have the capacity to really improve the living conditions of so many patients and their families.