|
Finding cell assemblies in brain activity| old_uid | 16450 |
|---|
| title | Finding cell assemblies in brain activity |
|---|
| start_date | 2018/09/27 |
|---|
| schedule | 14h |
|---|
| online | no |
|---|
| summary | The brain encodes information in the activity of “cell assemblies”, groups of neurons that are tied together by synaptic plasticity and are likely to activate in a synchronous way. In turn, cell assemblies activate in sequences, reflecting the temporal ordering of the events forming e.g. the memory of an episode. Spontaneous activity (taking place as the subject is inactive) is highly structured, and contains the activation of many cell assemblies, which may reflect stored memories, imagery, or planning of future actions. I present two methods for finding cell assemblies: In the first, we reconstruct the “functional connectivity matrix” by mapping recorded neural data on an a spin-glass network, and infer the maximum entropy model, in what is known as “reverse Ising inference”. From the connectivity matrix, cell assemblies can be reconstructed and their activity analyzed.
The second method tackles directly the temporal dimension by defining a distance between spike patterns inspired to the Earth Mover’s distance from Optimal Transportation theory, and then applying density-based clustering on the resulting distance matrix. Application of the method to simulated and real data reconstructs the structure of data and of the behavioral circumstances the animal experiences, in a completely unsupervised fashion.
I will frame these method in the context of systems neuroscience research, with particular focus on the study of memory systems. |
|---|
| responsibles | Taverna |
|---|
| |
|