Bridging the gap between cell types and spike trains
COSYNE 2026 Workshop
Understanding the brain requires connecting the language of computation—spike trains—to the biological reality of distinct neuronal cell types. Each cell type has characteristic connectivity, intrinsic dynamics, and receptor expression, which together shape its contribution to circuit function. Yet in most systems neuroscience experiments, this critical link remains hidden: we record spikes, but rarely know which cell types they come from.
To move from descriptive surveys of neural activity to genuine mechanistic understanding of brain function, we must characterize population dynamics in terms of interactions between distinct cell types. Advances in electrode technology now enable brain-wide recordings with thousands of neurons, yet the ability to assign these signals to defined cell types remains a major bottleneck.
Ongoing efforts are expanding optotagging-based cell type identification across more brain regions and species, revealing how distinct cell types contribute to network dynamics during behavior. In tandem, machine learning classifiers are beginning to enable cell type identification directly from electrophysiological features. A concerted push is needed to transform proof-of-concept models into robust, widely accessible tools for the community.
This workshop will serve as a catalyst for comparing current approaches to cell type inference and establishing a roadmap for using cell type identification to generate new insights into brain function. We expect it to appeal to algorithm developers, experimentalists, and theoreticians alike.
| Time | Speaker | Topic |
|---|---|---|
| 9:15 – 9:30 | Josh Siegle & Kenji Lee | Workshop introduction |
| 9:30 – 10:00 | Liset de la Prida | Cell-type specific decomposition of hippocampal manifolds See her new piece in The Transmitter! |
| 10:00 – 10:30 | Aditi Jha | Disentangling the roles of distinct cell classes with cell-type dynamical systems |
| 10:30 – 11:00 | Coffee break | |
| 11:00 – 11:30 | Maxime Beau | A deep learning strategy to identify cell types from high-density extracellular recordings |
| 11:30 – 12:00 | Manuel Valero | Cell-Type Rules of Hippocampal Coding |
| 12:00 – 12:30 | Anna Lakunina | Cell-Type-Resolved Electrophysiology With Neuropixels Opto |
| 12:30 – 3:00 | Lunch break | |
| 3:00 – 3:30 | Mehdi Azabou | Multi-animal models are good cell-type learners |
| 3:30 – 4:00 | Eva Dyer | Learning Who’s Who in the Brain: Contrastive Neuron-Level Embeddings for Cell Type Discovery |
| 4:00 – 4:30 | Panel discussion | Cross-species generalization, ground truth data, and building user-friendly tools |