Thèse Construction Cérébrale de la Perception de Mouvement Réel et Virtuel des Bases pour Mieux Vieillir H/F Doctorat.Gouv.Fr

Établissement : Université de Toulouse École doctorale : CLESCO - Comportement, Langage, Éducation, Socialisation, Cognition Laboratoire de recherche : CERCO - Centre de Recherche Cerveau et Cognition Direction de la thèse : Alexandra SEVERAC CAUQUIL ORCID 0000000332495741 Début de la thèse : 2027-09-01 Date limite de candidature : 2026-11-16T23:59:59 La connaissance de la position et du mouvement de notre corps et de ses segments dans l'espace est essentielle au contrôle de la posture et du percept de mouvement de soi tout au long de la vie. B-RealAge vise à identifier les réseaux cérébraux de la perception du mouvement de soi dans des contextes naturels et artificiels, en décrivant le traitement multisensoriel de l'espace chez l'humain. Nous envisageons d'aborder cette question en : (1) comprenant comment le cerveau intègre les différentes informations sensorielles pour élaborer une perception unique et cohérente du mouvement ; (2) évaluant l'influence du vieillissement sur ces processus ; et (3) explorant ces processus au-delà des conditions de base à l'aide de simulateurs et de dispositifs de réalité virtuelle. Nos hypothèses sont les suivantes : (1) les informations sensorielles diverses convergent dans le cerveau au sein de réseaux dédiés pour construire une perception unique et cohérente du mouvement de soi ; (2) les contextes artificiels, tels que la microgravité simulée et la réalité virtuelle, modifient l'équilibre du traitement du mouvement de soi ; et (3) le vieillissement interagit avec ces processus. Nos participants seront exposés à différentes situations de locomotion : marche libre ou expérience de mouvement de soi en position assise grâce à des dispositifs de réalité virtuelle ou à une stimulation sensorielle artificielle. Étant donné la complexité de l'architecture du réseau et sa possible variabilité selon les entrées sensorielles et les individus, nous élaborerons des modèles intégrant de multiples facteurs tels que la variabilité individuelle, l'âge et le sexe, en utilisant des approches de pointe incluant un modèle bayésien hiérarchique et l'apprentissage automatique. La maîtrise de ces processus permettra de mieux appréhender les risques et les écueils liés à l'utilisation de nouvelles technologies et de nouveaux dispositifs comme la réalité virtuelle et les simulateurs immersifs, tout au long de la vie et plus particulièrement lors du vieillissement. Crucial for self-motion processing, the integration of visual and vestibular input, known to begin as early in the central nervous system as in the brain stem (Angelaki & Cullen 2008) has been investigated at the cortical level in former studies including in our team (Aedo-Jury et al., 2020; Marchand et al., 2025). In particular, we described how within this network the cortical area V6 was the keystone of forward self-motion deciphering from visual (optic flow) and vestibular (GVS) cues .
Figure 1. Left panel: 4 stimulation conditions where visual and vestibular indicate forward or backward self-motion, in congruent or incongruent fashion. Right panel: mean+IC90 (n= 15) contrast value, expressed in arbitrary units (a.u.) within each ROI (shown in lateral and medial views of inflated brain) initially defined in our princeps study (Aedo-Jury et al. 2020). Contrast values for the Forward> Backward contrast. One-sample t-tests (** p<0.01).
The extensive use of novel technologies, in education, entertainment and rehab reveals side-effects and drawbacks such as headaches, cybersickness, etc, that could prevent from the intended, promising, goals. On top of these immediate effects, long-term repercussions are unknown and deserve investigation. The network architecture, which dynamically changes in response to the sensory input, is complex and may include individual difference. It is therefore necessary to use artificial intelligence for building the model including multiple factors such as individual variability, aging, and sex. The AIST group has constructed the pipeline using hierarchical Bayesian model and machine learning to detect the key network that is variable depending on the age and dynamically changes during performing the cognitive inhibition (Ouchi et al., 2026, preprint). This method will be used to achieve the purposes.
Knowing the position and motion of our body and its segments in space is decisive for the control of posture and self-motion during the whole lifespan. B-RealAge aims at describing the brain networks of self-motion perception in natural and artificial context by unveiling the multisensory processing of space in humans. We intend to tackle this issue by (1) understanding how the brain integrates the different sensory input to compute a unique and consistent percept of motion, (2) assess how aging interacts with these processes and (3) challenging the processes beyond basic conditions using simulators and virtual reality devices. Our hypotheses are that (1) multi-channel proprioceptive input converge in the brain in dedicated networks to build a single and coherent percept of self-motion (2) artificial context such as simulated microgravity and virtual reality reshuffle the balance of self-motion processing and (3) aging interacts with these processes. Mastering these processes would help to deal with risks and pitfalls of the use of novel technologies and devices such as virtual reality and immersive simulators. As the network architecture is complex and possibly varies with the sensory input and between individuals, we will build models including multiple factors such as individual variability, aging, and sex using state-of-the art approach. To address these 3 ambitious objectives, B-RealAge contains 4 work packages, that will be launched successively. (1) B-Real networks. Sensory channels will be stimulated using adequate and specific approaches mastered by the team: artificial self-motion signals will be provided by optic flow for vision, galvanic stimulation for the vestibular apparatus (Marchand et al., 2025). (2) B-real natural Brain activations while walking freely will be performed in the locomotion room available in the lab, using the portable fNIRS system. In this wp both visual and vestibular input will come from the actual motion of the subject. (3) B-noReal. Specific and challenging sensory context will be obtained by using virtual reality devices that strongly induces self-motion illusion, or vection, with VR goggles, that will provide vivid self-motion sensation via the sole visual channel (4) B-Real model hierarchical Bayesian model and machine learning with AI assistance will be implemented to modeling the networks involved.
For wp1&3, human neuro-imagery will be performed owing to an extensive and complete panel: MRI, long algo owned, used and mastered by the 2 labs. The very recently acquired MEG will now add time high accuracy to spatial.
For all wp, NIRS sport 2 system will be used since it allows brain activity recording in all positions of the subject, in both static (sitting, standing, lying down) or moving (walking, running) conditions.
For wp4, The model of sensory integration with/without real stimulation will be computed from the obtained data in wp1,2&3. To include the individual variability and aging, the hierarchical Bayesian model will be used, based on the outstanding expertise of the partner lab (AIST). In this study, we will develop a model that uses dynamic network changes related to sensory integration as explanatory variables and additional covariates such as age, gender, and cognitive function test scores.