Plug-in for the brain

31.01.2019
Interview

Through his work on Brain-Computer-Interfaces (BCI), EPFL Professor José R. del Millán helps people with physical disabilities live more independently. In an interview with the GDI, the researcher discusses BCIs and whether they could be hacked.

José del R. Millán

GDI: How does a Brain Computer Interface (BCI) work?

Jose del R. Millán: In a brain-computer interface (BCI), neural signals from the brain are fed into a decoding algorithm; this translates them into outputs so people with physical disabilities can control devices, from computer applications to exoskeletons. Feedback from the device–conveyed to the user via normal sensory pathways or through brain stimulation–establishes a closed control loop. BCI technology offers a natural way to augment human capabilities by providing a new link with the outside world. In this respect, it is particularly valuable for patients with severe neuromuscular disabilities.

What exiting applications of BCIs are coming, other than rehabilitation and patient care?

As the BCI field advances, the time is ripe to design interaction modalities for the average person. The goal is not to extend normal motor or sensory human capabilities (e.g., providing control of a third arm or infrared vision). Rather, BCI will predict actions the user will do (or not do,) and decode the user’s cognitive state. This will allow the intelligent device to assist the user. An example of this emerging research is the use of BCIs for enhancing car driving.

What is the biggest obstacle you have to overcome in order to make brain computer interfaces work smoothly so they can be implemented for the mass market?

Clearly, we need to increase the level of performance of a BCI for deploying it in the mass market. But this is not the main priority as, for most BCI users, the current level of performance should be sufficient to enhance interaction experiences in non-critical applications. To me, the major barriers are three. First, given the nature of brain signals that are highly variable, a BCI cannot guarantee good decoding of users' intentions all the time. We need to find ways to cope with variability in performance. Second, setting up a BCI must be simplified so only a minimal level of expertise is required to operate. Third, current brain recording systems are too obtrusive and they should become more transparent.

Will we ever have BCIs that are working with the mere power of thought and without electrodes?

This will be difficult. The reason is that brain signals (either electrical, chemical or of any other nature) are very tiny and recording devices must have high sensitivity to record them with sufficient quality (signal-to-noise ratio) over other surrounding signals inside and outside our body.

Can you manipulate people by using BCIs?

Users need to learn to generate specific brain patterns in order to operate their BCI. This means one could interfere with normal brain processes that give rise to the target patterns, thus impeding BCI usage. This is relatively easy to do, as it is similar to what would happen if someone disturbed you while doing a manual task. Hypothetically, we could induce a stronger level of manipulation by externally generating in BCI users' brain the same patterns that they elicit to operate their BCI, making them believe they are voluntarily engaging in the interaction.

José del R. Millán is "Defitech" professor and head of the laboratory for brain-machine interfaces at EPFL. He is a speaker at the 15th European Trend Day, which will take place on 13 March 2019 at the Gottlieb Duttweiler Institute. The focus will be on the future of communication, which internationally renowned thought leaders will discuss.