Department of Brain and Cognitive Sciences
Collective Behavior and Scaling in (very) Large Populations of Neurons
Recent technological progress has dramatically increased our access to the neural activity underlying memory-related tasks. These complex high-dimensional data call for theories that allow us to identify signatures of collective activity in the networks that are crucial for the emergence of cognitive functions. As an example, we study the simultaneous activity of ~2000 neurons imaged in dorsal hippocampus as a mouse runs along a virtual linear track. To bridge the scales between individual neurons to population-level computations, we use coarse-graining methods inspired by the concept of renormalization group (RG). The RG formalizes how macroscopic theories can be simpler and more universal than their underlying microscopic mechanisms. When coarse-graining in the RG spirit, we uncover macroscopic features of the network, and follow the changes in coding properties as we scale up. We see nearly-perfect scaling in both static and dynamic properties of the activity, in addition to hints of it being controlled by a non-trivial fixed point. Perhaps, then, there is emergent simplicity even in neural activity as complex as rodent hippocampus.
A pizza lunch will be served.