Homepage Inria website

Section: New Results

Improving User-Training for Brain-Computer Interfaces

Participants : Martin Hachet, Emilie Jahanpour, Camille Jeunet, Fabien Lotte, Boris Masencal, Julia Schumacher.

While Mental Imagery based Brain-Computer Interfaces (MI-BCIs) are promising for many applications, ranging from assistive technologies for motor disabled patients to video games, their usability“out-of-the-lab” has been questioned due to their lack of reliability: literature reports that 15% to 30% of users cannot control such a technology, while most of the remaining users obtain only modest performances. As controlling an MI-BCI requires the acquisition of specific skills (i.e., producing stable and distinct brain-activity patterns), an adapted training is necessary. Thus, the main objective of our project is to improve the user training to facilitate the acquisition of MI-BCI related skills. In order to do so, we focused on two axes [18] : (1) the impact of the user-profile and (2) the impact of the protocol on MI-BCI performance.

Concerning the impact of the user-profile, our results ([40] , [14] ) suggested an important impact of some aspects of the personality (such as the tension and autonomy levels) as the spatial abilities (i.e., the ability to produce, interpret and transform mental imageries). On the one hand, we are working on learning companions, whose goal would be to provide the learners with a specific emotional support, based on their profile and on their cognitive state. On the other hand, we are currently implementing and testing a spatial ability training in order to test the hypothesis of a causal effect of the spatial abilities on MI-BCI performance [39] . In other words, we would like to know if increasing spatial abilities would result in better MI-BCI performance. One application of such a research is stroke rehabilitation. Indeed, motor after-effects are usual following a stroke. MI-BCI have been shown very useful to facilitate the rehabilitation process, which consists in enhancing brain plasticity through motor-imagery, as they enable to visualise the BCI activity while the patients perform MI-tasks. However, MI-tasks tend to increase the depressive state of the patients as they remind them they lost the ability to move their limb. Thus, as spatial ability exercises (e.g., mental rotation) activate the motor cortex, they could be used as more transparent rehabilitation exercises to trigger brain plasticity.

Second, concerning the impact of the protocol, we completed a study (see activity report 2014) in which we asked the participants to use the standard MI-BCI training protocol to learn to perform simple motor tasks: drawing circles and triangles on a graphic tablet. As it would have been the case for an MI-BCI experiment, they had to find the right strategy so that the system recognises the task they were performing. Seventeen percent of the participants (N=54) showed difficulties in performing these tasks. Also, when we selected the 10 best and 10 worst performers of this experiment and asked them to use an MI-BCI (by imagining left and right-han movements), it appeared that the ones who had difficulties in performing the simple motor tasks improved in terms of performance during the MI-BCI experiment, while the participants who performed well during the motor experiment did not progress during the second. Furthermore, we have shown that tactile feedback was more efficient than an equivalent visual feedback in a multitasking context [32] . Based on a literature review, this could be due to an increased sense of agency (i.e., the feeling to be in control). We are thus currently exploring the impact of the sense of agency on MI-BCI performance. Finally, still regarding the feedback, we explored what kind of information could help the user to perform better mental imagery tasks. As such, we look for physiological features that could predict whether a mental task will be correctly recognized by the BCI, and that could be understood by the user. Among the different features we explored, it appears that the user's relaxation (from a muscular point of view), as measured in EMG activity collected by EEG channels, is one of such features. We are currently building and exploring new BCI training protocols that provide additional information about the user's muscular relaxation as complementary feedback [34] .