Neuroscience robotics for controlled induction and real-time assessment of hallucinations

Although hallucinations are important and frequent symptoms in major psychiatric and neurological diseases, little is known about their brain mechanisms. Hallucinations are unpredictable and private experiences, making their investigation, quantification and assessment highly challenging. A major shortcoming in hallucination research is the absence of methods able to induce specific and short-lasting hallucinations, which resemble clinical hallucinations, can be elicited repeatedly and vary across experimental conditions. By integrating clinical observations and recent advances in cognitive neuroscience with robotics, we have designed a novel device and sensorimotor method able to repeatedly induce a specific, clinically relevant hallucination: presence hallucination. Presence hallucinations are induced by applying specific conflicting (spatiotemporal) sensorimotor stimulation including an upper extremity and the torso of the participant. Another, MRI-compatible, robotic device using similar sensorimotor stimulation permitted the identification of the brain mechanisms of these hallucinations. Enabling the identification of behavioral and a frontotemporal neural biomarkers of hallucinations, under fully controlled experimental conditions and in real-time, this method can be applied in healthy participants as well as patients with schizophrenia, neurodegenerative disease or other hallucinations. The execution of these protocols requires intermediate-level skills in cognitive neuroscience and MRI processing, as well as minimal coding experience to control the robotic device. These protocols take ~3 h to be completed.

Hallucination Engineering: From Neuroscience Robotics to Hallucinations in Health and Disease

Eva Blondiaux

My thesis focuses on a psychotic symptom called the presence hallucination (PH). PHs are defined as the false perception that someone is nearby when no one is actually present. PHs can occur in various populations, ranging from healthy subjects (when exposed to extreme conditions) to neurological (e.g. epileptic, Parkinsonâs disease patients) and psychiatric patients (e.g. schizophrenia). Our group has designed a robotic device that allows to safely induce PH in healthy subjects, in a controlled manner, using different sensorimotor conflicts (Blanke et al., 2014). The aim of my thesis was to unravel the brain regions and mechanisms associated with PH and explore how the neuroscientific understanding of PH can improve diagnostics of different clinical populations with hallucinations. In Chapter 2, I investigated the brain networks associated with a category of complex phenomena named autoscopic phenomena (AP). AP include PHs and three other forms: autoscopic hallucinations, heautoscopies, and out-of-body experiences. AP are rare phenomena, defined as illusory reduplication involving a double of oneâs own body seen or felt in extrapersonal space. Lesion network mapping analysis was used to identify the brain networks associated to each AP. My data showed that each AP is characterised by specific connectivity networks consistent with their phenomenology. All AP also share common brain connectivity. In Chapter 3, using an MR-compatible robot able to induce PH, I investigated the brain regions and mechanisms of robot-induced PH in healthy subjects using fMRI. I combined these data with the results of my lesion network mapping analysis in neurological patients with focal brain damage associated with PH (Chapter 3). Both networks (i.e. robot-induced PH-network in healthy subjects and symptomatic PH-network) were merged together to identify the key regions involved in PH, defining a common PH-network. The relevance of this common PH-network was then assessed and confirmed in patients with Parkinsonâs disease (PD), revealing a difference in functional connectivity within this network between PD patients experiencing symptomatic PH compared to patients without such symptoms. Further, I studied the functional connectivity within the common PH-network in 22q11.2 deletion syndrome (DS) subjects (i.e. subjects with high risks (30%) of developing schizophrenia) (Chapter 4) and in psychotic patients with passivity experiences (Chapter 5). Studying PH in these population is relevant since PH is frequent in schizophrenia. In both populations, a reduced functional connectivity in the common PH-network was revealed. In Chapter 4, I also assessed the sensitivity to the sensorimotor conflicts and the proneness to experience robot-induced PH in 22q11DS subjects and age-matched controls. Results showed a lack of sensitivity in sensorimotor modulation for the sense of agency, further corroborated by a lack of delay dependency in experiencing robot-induced PH in 22q11DS subjects compared to controls. In summary, by coupling robotic technology with neuroimaging, a common PH-network was delineated and was found relevant in different clinical populations. These findings are important since PH can be present at early stages before the apparition of more severe symptoms. Therefore, our results can lead to more robust diagnostic tools and early treatment intervention, which might significantly improve prognosis in those patients.

EPFL2020

Hallucination Engineering: From Neuroscience Robotics to Hallucinations in Health and Disease

Eva Blondiaux

EPFL2020