Brain-computer interfaces or control systems by brain activities

Tell us about your career so far ?

I started out as a computer scientist. After graduating as a computer engineer and completing a Master’s research degree in Artificial Intelligence and Imaging at INSA and the University of Rennes, I did a PhD in computer science at IRISA/Inria Rennes and, finally, a postdoc at the Institute for Infocomm Research (I2R) in Singapore. I have already worked for six years as a researcher at Inria Bordeaux as part of the POTIOC team, which is mainly concerned with designing, developing and assessing new human-machine interaction methods, including brain-computer interfaces.

What is your research about?

My research focuses on brain-computer interfaces, or BCIs, which are command and control systems allowing human users to send commands to a computer using only their brain activity, which is generally measured by an electroencephalograph (EEG) and processed by the system. In BCIs, the EEG uses electrodes placed on the scalp to measure electrical microcurrents that reflect the synchronised activity of millions of neurons in our brain.  
What I’m particularly interested in is how EEG signals can be analysed and processed to identify as accurately as possible the mental command that the user wishes to send. I also do research on how to optimise the learning process for BCI users. We already know that controlling a BCI is like riding a bicycle: first you must learn how to do it, then the more you practice, the better you become. Lastly, I try to think up new applications for brain-computer interfaces.

What do you find so fascinating about this field of research?

A lot of things! First of all, it’s a new field of research, which makes it rich and dynamic. There are many themes to explore and new projects to invent. In addition, BCIs are at the meeting point of numerous disciplines, such as computer science, mathematics, signal processing, neuroscience, psychology and medical research. That means I am always on the look-out for new topics, which is intellectually very stimulating. And I must admit that the sci-fi side of BCIs also particularly appealed to me.

What’s the theme of the project for which you were awarded an ERC Starting Grant?

As I pointed out earlier, BCI research is still in its early stages, which implies a high rate of error (one in four on average) in use. My project, called Brain Conquest: Boosting Brain-Computer Communication with High Quality User Training sets out to enhance the reliability of BCIs. Solving this problem means studying the users’ ability to control the system. That’s why I’m particularly interested in the users’ learning process when they try to control a BCI. 
To do that, I’m going to try and create learning process models, based on experimental results and psychological, mathematical or computational models, in order to understand more about the resulting cognitive and mental workings. These studies seek to automatically guide users as effectively as possible as they learn to control a BCI. Eventually, I would like to develop BCIs that people with disabilities could use to control assistive systems (such as controlling a text editor using only their brain) or, in a completely different area, video games - serious games in particular - controlled by BCIs.

What makes your approach original?

Until now, most researchers in the field - myself included - have focused on analysing, processing and classifying EEG signals in order to recognise the mental command that the user wishes to execute. Although these methods have improved BCIs, performance levels are still too low. So we need to explore new avenues! Research to date has given little or no consideration to the user, the human aspect of the problem.

So rather than working to perfect EEG signal processing, what I’d like to do now is improve the way in which humans use BCIs.

We have already shown that current learning methods for controlling BCIs are less than optimal, in both practical and theoretical terms. My approach is original in that it seeks to develop new training methods that will incorporate the user’s profile. There is already some research in progress on human learning, but the approach adopted is experimental and imperfect. I wish to develop models and theories that will provide a basis for creating relevant methods for learning to control BCIs.

In concrete terms, how will this grant help you in your research?

To start with, I wish to hire three doctoral candidates, two postdocs, an engineer and several interns from various backgrounds to form the type of interdisciplinary team needed to carry out this project, which touches upon so many different fields (human-machine interface, signal processing, machine learning and cognitive science). The grant will also help to purchase high-resolution, and therefore more accurate, EEG signal measuring equipment. Lastly, I’ll do my best to promote exchanges with researchers from all around the world to share our experience and expertise in this field.

What’s the formula for obtaining this prestigious grant?

I don’t think there’s any magic formula, but I would say that the people you work with are the key. An ERC is a goal you work towards by exchanging, talking, sharing - in my case with the POTIOC team and other experts in the field. The application for an ERC has to be set up with support from Inria’s European project leaders. And an ERC has to be prepared based on Jean-Pierre Banâtre’s precious advice (former Director of European Partnerships). There’s another essential ingredient for landing this grant, and that was many hours of work, especially in writing up the project, where every word was carefully weighed. So I would say that I owe this grant to my own work but also, and above all, to all the people who worked with me along the way, whether near or far. I would like to sincerely thank them.