GSO-2010: John Beggs
Information flow in networks of 100 cortical neurons is non-democratic
The average cortical neuron makes and receives about 7,000 synaptic contacts. This anatomical information suggests that all neurons should participate equally in processing information. But the physical connections found in the brain do not necessarily reveal how information flows there. We used a new measure called transfer entropy to assess information flow in small samples of cortical tissue placed on a 512 electrode array system. Our analysis revealed that, far from being democratic, information flow in these networks was distributed in highly uneven manner. About 80% of the total information flow passed through only 20% of the neurons. Simulations indicated that this non-democratic pattern of information flow can actually increase the efficiency and robustness of the network.
Work funded by NSF and done in collaboration with Shinya Ito, Aonan Tang, Alan Litke and Olaf Sporns.