Back to overview

Dissecting the neuronal circuit mechanisms underlying neuroprosthetic learning

English title Dissecting the neuronal circuit mechanisms underlying neuroprosthetic learning
Applicant Huber Daniel
Number 184829
Funding scheme Project funding (Div. I-III)
Research institution Dépt des Neurosciences Fondamentales Faculté de Médecine Université de Genève
Institution of higher education University of Geneva - GE
Main discipline Neurophysiology and Brain Research
Start/End 01.06.2019 - 31.05.2023
Approved amount 846'720.00
Show all

Keywords (6)

Neuroprosthetic Learning; Feedback; Brain machine interface; Two-Photon Microscopy; Systems Neurscience; Cortex

Lay Summary (German)

Wie passt sich unser Gehirn einer "Brain-Machine Interface" an?
Lay summary

Brain-machine interfaces (BMIs) stellen künstliche Verbindungen zwischen der Umwelt und dem Gehirn her. Sie versprechen daher wichtige klinische Anwendungen, indem sie möglicherweise ausgewählte Aspekte sensorischer oder motorischer Funktionen ersetzen, wiederherstellen oder erweitern können. BMIs sind darauf angewiesen, neuronale Signale aus dem Gehirn auszulesen, um externe Glieder zu bewegen, oder um Informationen von externen Sensoren an das Gehirn zurückzuleiten oder beides gleichzeitig. Dieses künstliche Assoziation muss erlernt werden und dieser Prozess wird als neuroprothetisches Lernen bezeichnet wird. Zunehmende Daten deutet darauf hin, dass die Plastizitätsmechanismen die dem neuroprothetischen Lernen zugrunde liegen ähnlich verlaufen wie beim natürlichen Lernen. Da das neuroprothetische Lernen die Komplexität umgeht, die mit natürlichen sensorischen Eindrücken oder motorischen Kontrolle verbunden ist, können die beteiligten Schaltkreise leichter identifiziert und charakterisiert werden. Diese Art von Forschung hat daher wichtige Vorteile gegenüber natürlichem Lernen und bietet eine interessante Alternative zur Analyse neuronaler Schaltkreise, die dem sensorisch-motorischen Lernen zugrunde liegen. Wir haben kürzlich die Machbarkeit einer rein optischen BMI demonstriert, indem wir die Aktivität identifizierter kortikaler Neuronen auf das Feedback einer künstlichen sensorischen Eingabe konditionieren. Diese Experimente gaben einen ersten Einblick in die Aktivitätsdynamik der konditionierten Neuronen sowie ihres umgebenden Netzwerks während des Lernens. Das übergeordnete Ziel dieses Projekts ist es, die grundlegenden Mechanismen des neuroprothetischen Lernens von der Zelle bis zur Schaltkreisebene besser zu verstehen.

Direct link to Lay Summary Last update: 14.05.2019

Responsible applicant and co-applicants


Associated projects

Number Title Start Funding scheme
163762 Optical Imaging of Cortical Motor Control 01.01.2016 SNSF Professorships


Background: Brain-machine interfaces (BMIs) create artificial links between the environment and brain circuits. They therefore promise important clinical applications by potentially replacing, restoring or augmenting selected aspects of sensory or motor functions. BMIs rely on reading out neuronal signals from the brain to move external actuators or are designed to feed information from external sensors back to the brain or carry out both simultaneously in a closed loop. This imposed artificial input-output mapping needs to be integrated or learned in a process which is termed neuroprosthetic learning. Increasing evidence suggests that plasticity mechanisms underlying neuroprosthetic learning are closely related to the ones involved in forming natural sensory-motor associations. Because neuroprosthetic learning circumvents the complexity associated with natural sensory input or motor output, the involved circuits can be more easily identified and characterized. From a research perspective, this behavioral framework has therefore important advantages over more natural learning paradigms and provides a powerful alternative for dissecting neuronal circuits underlying sensory-motor learning. We have recently demonstrated the feasibility of an all-optical closed-loop version of a BMI, by conditioning the activity of identified cortical neurons to the feedback of an artificial sensory input. These experiments provided a first glimpse at the activity dynamics of the conditioned neurons as well as their surrounding network during learning. The overarching goal of this project is to gain a better understanding of the basic mechanisms underlying neuroprosthetic learning from the cellular to the circuits level, by building on our expertise in brain-machine interfaces, chronic two-photon microscopy and optogenetics in awake mice. Specifically we propose to: •Aim 1: Characterize the structural correlates of plasticity induced by neuroprosthetic learning. By combining virus based anatomical single-cell reconstruction and optical imaging methods, we will investigate the origin and relevance of afferent inputs that drive the conditioned neuronal activity. In parallel, we will dissect the synaptic plasticity mechanisms involved in the learning process by tracking changes in dendritic structures that are driven by the input activity. Together, these experiments will allow us to determine if learning occurs locally, as predicted by several recent studies, and help us define the basic circuit elements and areas involved in the synaptic changes that lead to neuroprosthetic learning. •Aim 2: Probe the cellular factors affecting the long term stability of neuroprosthetic learning. Using single cell conditioning of different cell types located in different cortical layers, we will systematically test which factors impact the long term stability of neuronal changes induced by neuroprosthetic learning. In addition, we will study the role of sleep in the long term consolidation process by characterizing the activity of the trained neuron and the related afferent input during post-conditioning sleep. •Aim 3: Test the limits of artificial feedback channels used for neuroprosthetic learning. Finally, we will build on our previous experience using optogenetic cortical microstimulation in behaving mice to explore the limits of artificial sensory signals to provide feedback. These signals are crucial to generate relevant and informative artificial feedback during neuroprosthetic learning. The results of these experiments might potentially improve the design and coding of artificial feedback in future closed-loop BMIs. These three aims will be carried out by two postdocs and one graduate student who will interact at multiple levels over the duration of 4 years. The majority of the necessary imaging equipment, optogenetic tools and animal lines are already in place in my laboratory. Expected value: We expect that this project will advance our understanding of the neuronal mechanisms involved in neuroprosthetic learning and thereby not only reveal some of the most basic cortical plasticity mechanisms underlying sensory-motor learning, but in parallel lay the groundwork for improving future clinical implementations of BMIs.