PhD Position F/M Neuroadaptive neurofeedback user training (NUT)

The Inria Centre at Rennes University is one of Inria's eight centres and has more than thirty research teams. The Inria Centre is a major and recognized player in the field of digital sciences. It is at the heart of a rich R&D and innovation ecosystem: highly innovative PMEs, large industrial groups, competitiveness clusters, research and higher education players, laboratories of excellence, technological research institute, etc.

This PhD is not in the context of a funded partnership. However, a collaboration with Marius Klug from Brandenburgische Technische Universität Cottbus-Senftenberg is envisioned.

The goal of the project is to develop innovative and neuroadaptative neurofeedback interfaces. During neurofeedback training, people are provided with direct feedback regarding their own brain activity in order to learn to control it, often with the aim to improve their cognitive abilities. Previous research demonstrated that the state of neurofeedback trainees, such as their attentional state (workload, attention, motivation,...) influences their ability to learn to control their brain activity. Thus, the goal of this PhD is to adapt the neurofeedback training to better account for the state of the trainee.

No regular travel is foreseen for this position, but a mobility to a foreign research laboratory or industry will be encouraged.

Keywords: Neurofeedback, modeling, adaptive tutoring

PhD topic: Controlling one's own brain activity when receiving direct information regarding the former is a skill that can be acquired using neurofeedback training. During such training, people's brain activity is acquired, often using electroencephalography (EEG), and converted into a feedback that people have to learn to control [Roc et al., 2021]. The ability to modulate one's own brain activity can be used for two main types of applications. First, to use brain-computer interfaces (BCIs), that enable the control of external digital systems by producing discriminatory and stable brain patterns each associated with a specific command for the system [Wolpaw et al., 2012 ; Roc et al., 2021]. For instance, BCIs can be used to control the direction of a character in a video game or the direction of a wheelchair by imagining right or left-hand movements [Tonin et al., 2022]. Second, for neurofeedback applications for which the end goal is that the modifications occurring in the brain activity lead to cognitive improvements, often in clinical applications [Batail et al., 2019]. For instance, neurofeedback can be used for motor rehabilitation after a stroke [Le Franc et al., 2022].

The state of a person is defined as his/her characteristics that are “temporary, brief, and caused by external circumstances”. It is already known that mental states, such as attention or workload, influence the ability to control one's own brain activity [Kadosh and Staunton, 2019 ; Tzdaka et al., 2021]. However, very few studies adapted the training to take into account the state of the user into account. One of the very first study to do so is the one from Myrden and Chau from 2016 during which they used self-reported levels of fatigue, frustration and attention to adapt a mental task neurofeedback user training. In another study, Talukdar et al., 2020, developed a method to potentially take into account the state of fatigue of the participants into the signal processing method, without testing the effect of such adaptation. In any type of training, the state of the learners are decisive in the outcome of the training [Keller and Keller, 2010]. Numerous types of applications are already taking advantage of these pieces of information, such as health [Jovanov et al., 2005], sport [Baca and Kornfeind, 2006] or intelligent tutoring systems [Woolf et al., 2010]. Assessing the states, including attention, working memory, emotions or motivation, of the users would thus be relevant to improve neurofeedback learning as well. Assessing the state of the users would not require further equipment as it could already be inferred using the neurophysiological data acquired for neurofeedback training.

Such adaptation could concern several main characteristics of the training, including the content, modality and timing of the feedback or the type of mental tasks that people are asked to perform. For instance, the number of mental tasks could be reduced or the type of mental tasks could be made easier when an increase in workload is detected in the neurophysiological data of the user. Another possibility would be to enrich the neurofeedback provided to the users to convey additional information related to their mental state. Few studies have been led in order to enrich the traditional evaluative feedback. For instance, Sollfrank et al. chose to add information concerning the stability of the EEG signals to the standard feedback based on classification accuracy [Sollfrank et al., 2016], while Schumacher et al. added an explanatory feedback based on the level of muscular relaxation to this classification accuracy-based feedback [Schumacher et al., 2015]. To our knowledge, none have provided feedback on the mental state of the learners during neurofeedback training.

On a fundamental level, this thesis will provide a better understanding of the neuromarkers of mental states as well as first computational models on their influence on neurofeedback performances. Experiments will make it possible to test, improve and validate the models, methods and algorithms created.

References:

• Baca, A., & Kornfeind, P. (2006). Rapid feedback systems for elite sports training. IEEE Pervasive Computing, 5(4), 70-76.
• Batail, J. M., Bioulac, S., Cabestaing, F., Daudet, C., Drapier, D., Fouillen, M., ... & Vialatte, F. (2019). EEG neurofeedback research: A fertile ground for psychiatry?. L'encephale, 45(3), 245-255.
• Jovanov, E., Milenkovic, A., Otto, C., & De Groen, P. C. (2005). A wireless body area network of intelligent motion sensors for computer assisted physical rehabilitation. Journal of NeuroEngineering and rehabilitation, 2, 1-10.
• Kadosh, K. C., & Staunton, G. (2019). A systematic review of the psychological factors that influence neurofeedback learning outcomes. Neuroimage, 185, 545-555.
• Keller, J. M., & Keller, J. M. (2010). The Arcs model of motivational design. Motivational design for learning and performance: The ARCS model approach, 43-74.
• Le Franc, S., Herrera Altamira, G., Guillen, M., Butet, S., Fleck, S., Lécuyer, A., ... & Bonan, I. (2022). Toward an adapted neurofeedback for post-stroke motor rehabilitation: state of the art and perspectives. Frontiers in Human Neuroscience, 16, 917909.
• Myrden, A., & Chau, T. (2016). Towards psychologically adaptive brain–computer interfaces. Journal of neural engineering, 13(6), 066022.
• Pillette, L., Appriou, A., Cichocki, A., N'Kaoua, B., & Lotte, F. (2018, May). Classification of attention types in EEG signals. In 7th International BCI Meeting.
• Roc, A., Pillette, L., Mladenovic, J., Benaroch, C., N’Kaoua, B., Jeunet, C., & Lotte, F. (2021). A review of user training methods in brain computer interfaces based on mental tasks. Journal of Neural Engineering, 18(1), 011002.
• Schumacher, J., Jeunet, C., & Lotte, F. (2015, October). Towards explanatory feedback for user training in brain-computer interfaces. In 2015 IEEE International Conference on Systems, Man, and Cybernetics (pp. 3169-3174). IEEE.
• Sollfrank, T., Ramsay, A., Perdikis, S., Williamson, J., Murray-Smith, R., Leeb, R., ... & Kübler, A. (2016). The effect of multimodal and enriched feedback on SMR-BCI performance. Clinical Neurophysiology, 127(1), 490-498.
• Talukdar, U., Hazarika, S. M., & Gan, J. Q. (2020). Adaptive feature extraction in EEG-based motor imagery BCI: Tracking mental fatigue. Journal of neural engineering, 17(1), 016020. 
• Tonin, L., Perdikis, S., Kuzu, T. D., Pardo, J., Orset, B., Lee, K., ... & Millán, J. D. R. (2022). Learning to control a BMI-driven wheelchair for people with severe tetraplegia. Iscience, 25(12).
• Tzdaka, E., Benaroch, C., Jeunet, C., & Lotte, F. (2020, October). Assessing the relevance of neurophysiological patterns to predict motor imagery-based BCI users’ performance. In 2020 IEEE International Conference on Systems, Man, and Cybernetics (SMC) (pp. 2490-2495). IEEE.
• Wolpaw, J. R., Millan, J. D. R., & Ramsey, N. F. (2020). Brain-computer interfaces: Definitions and principles. Handbook of clinical neurology, 168, 15-23.
• Woolf, B. P., Arroyo, I., Cooper, D., Burleson, W., & Muldner, K. (2010). Affective tutors: Automatic detection of and response to student emotion. Advances in intelligent tutoring systems, 207-227.

Collaboration: The PhD student recruited will be supervised by Léa Pillette (CNRS researcher in the Hybrid team at IRISA), Marc Macé (CNRS researcher in the Hybrid team at IRISA) and Anatole Lécuyer (Inria research director in the Hybrid team at IRISA). Marius Klug, young investigator and group leader from Brandenburgische Technische Universität Cottbus-Senftenberg will be involved in the co-supervision of the PhD student. Marius Klug’s expertise in the assessment of users’ state using physiological data, notably neurophysiological ones, and his expertise in developing engaging and immersive interfaces will be highly valuable in the supervision of the PhD student.

Main activities (5 maximum) :

• Literature review
• Experimental design
• Implementation of neurofeedback solutions
• Statistical and neurophysiological analyses
• Write scientific articles

Additional activities (3 maximum) :

• Present research results at conferences
• Potentially teaching classes
• Potential supervision of intern project

We are looking for a highly motivated student with advanced skills in computational neuroscience and a good level in English.

• Subsidized meals
• Partial reimbursement of public transport costs
• Possibility of teleworking (90 days per year) and flexible organization of working hours
• Partial payment of insurance costs

Monthly gross salary: 2100€ during the 2 1st years and 2200€ during the 3rd year.