Analog VLSI implementation of a nonlinear systems model of the hippocampal brain region

The hippocampus is a major brain system involved in learning and memory functions, and consists of multiple populations of neurons with strongly nonlinear properties that are interconnected both locally and non-locally. An analog VLSI design has been developed that allows different classes of nonlinearities specific to each neuron population to define the transfer function of a network of neurons implemented in hardware. Principles of a CNN design have been used to generate local interactions between adjacent processing elements. Non-local interactions will be implemented in future designs with the use of multiple chips. In this manner, we are attempting to better integrate into a hardware device the unique information processing and learning capabilities of real biological neurons known to perform those functions.<<ETX>>