Professor Elisha Moses
DEPARTMENT OF PHYSICS
Weizmann Institute of Science

Simple Computation Devices from Living Neural Networks

We examine some necessary conditions for neurons to carry out a computation. Using both 1D and 2D configurations of hippocampal cultures, we study living neural networks in ways that are inaccessible to standard electrophysiology. In 2D cultured networks, the use of percolation and of graph theory reveals the input connectivity matrix of the culture. In 1D, the basic transport of signals along an information channel can be examined. Focusing on the threshold function created by the need for multiple inputs, we show that the role of the integration in the 'integrate and fire' neuron is crucial for logic functions. Based on this, we are able to design and engineer hybrid 1D-2D geometries that form functional neuronal logic devices, using multiplexing and redundancy to create reliable computation.