Neuralink-like brain machine interface in Virtual Reality
Master thesis / Research Assistant project:
Investigating neural population dynamics across 12+ brain regions using a 1024-channel brain-machine interface during a VR learning task in rats
Master thesis / Research Assistant project:
Investigating neural population dynamics across 12+ brain regions using a 1024-channel brain-machine interface during a VR learning task in rats
Understanding how neural populations change their coding during learning is a central question in systems neuroscience[3]. This project aims to shed light on how changes in the low-dimensional subspaces of population activity across brain regions facilitate learning. [1]
We have established a behavioral paradigm in rats in which we observe the gradual emergence of a behavioral strategy. We then reverse the paradigm rules to probe the animals' ability to generalize.
[1] Averbeck, B., Latham, P. & Pouget, A. Neural correlations, population coding and computation. Nat Rev Neurosci 7, 358–366 (2006). https://doi.org/10.1038/nrn1888
[3] Sun, W., Winnubst, J., Natrajan, M. et al. Learning produces an orthogonalized state machine in the hippocampus. Nature 640, 165–175 (2025). https://doi.org/10.1038/s41586-024-08548-w
How do changes in behavioral strategy manifest in neural population coding? How do higher-order areas such as the medial prefrontal cortex (mPFC) differ from primary and secondary sensory cortices in terms of plasticity during learning and reversal learning?
Can we identify features of neural population dynamics that facilitate learning? And how are these features coupled to changes in behavior?
Custom flexible electrodes, high-density chronic BMI, and behaving rats in Virtual Reality
Learning moderately complex tasks unfolds over extended timescales. To observe how populations of neurons across multiple brain regions adapt, it is essential to track neural activity longitudinally. We are uniquely positioned using our flexible electrode technology, which enables long-term, stable recordings across brain regions [2].
Our latest brain-machine interface (BMI) system allows simultaneous acquisition of high-density neural data—up to 1408 electrodes spanning 12+ distinct brain areas. This cross-regional perspective is crucial for capturing the distributed nature of learning in the brain.
[2] Yasar, T.B., Gombkoto, P., Vyssotski, A.L. et al. Months-long tracking of neuronal ensembles spanning multiple brain areas with Ultra-Flexible Tentacle Electrodes. Nat Commun 15, 4822 (2024). https://doi.org/10.1038/s41467-024-49226-9
Join our research team and contribute to cutting-edge systems neuroscience
We have successfully established the behavioral paradigm and acquired two months of neural data from a single rat performing the task. For the next phase of the project, we are seeking a motivated candidate to join the team.
The candidate can focus on one of two complementary areas that can be tailored to your interests and background. The starting date is Summer 2025, and we prefer students that stay long (6+ months)
Highly qualified applicants may be eligible for compensation based on experience and project contribution.
LTK certificate, passionate to work with animals and cutting edge technology, very high attention to detail, patience with animals
Strong in Python, statistics, machine learning. Experience analyzing neural data, neuroscience domain knowledge, software development experience
Get in touch
Neurotechnology Group
ETH Zurich | UZH
Institute of Neuroinformatics
Zurich, Switzerland
