Théorie et applications de la dynamique entraînée de réseaux neuronaux récurrents du cerveau


Guillaume Lajoie

Université de Montréal


Domaine : neurosciences, santé mentale et toxicomanies

Programme Chercheurs-boursiers - Junior 1

Concours 2018-2019

Is it possible to read out information from neural circuits directly, and repair them when they are broken? With the use of mathematical models and simulations, my research program seeks answers to these questions, with specific contributions to the development of neural implants that directly interact with circuits in the brain.

Networks of neurons in the brain have important constraints: (1) they are susceptible to injuries, diseases, and aging, and (2) they are constrained by limitations of the peripheral nervous system — information about the world reaches our brain via our sense, and we interact with the world via movements of our muscles, both of which have limited speed and bandwidths. Is it possible to use neural implants that interact directly with the brain to (1) mitigate the effects of pathologies, and (2) provide novel input/output modalities between our brains and the world?

Brain-Computer Interfaces (BCI), are designed to do just that. More specifically Bidirectional BCIs (BBCI), implants that not only allow the reading out of neural activity, but also the delivery of electrical signals to neurons, enable a two-way interaction with the brain. These devices not only influence the spiking activity of neurons, but can impact the plasticity mechanisms that modulate the synaptic connections between them, thus affording the possibility to "engineer" new circuits in the brain. As BBCI technology develops, the capacity to record and stimulate more and more neurons grows, and for such bidirectional neural implants to be successful, a number of real-time computational issues need to be resolved.

As an applied mathematician, I will work closely with neuroscientists, clinicians and engineers to implement computational tools that address these issues.