STATISTICAL MECHANICS OF NEURAL NETWORKS,
Abstract
A mathematical model is developed of the activity of networks of model nerve-cells or neurons. A nonlinear delay operator is introduced to represent the transfer characteristics of neurons. This operator is a continuous function that represents the mean neuronal response to stimulating currents. The mathematics used is not the Boolean algebra of switching circuits, but Differential equations. The dynamics is examined of certain model brain circuits, some of which are shown to exhibit undamped responses to stimuli, but only for a few levels of activity. The techniques of Hamiltonian mechanics and of Gibbsian statistical mechanics are used to connect these models with experimental data. A preliminary explanation is given for the existence of preferred states found in the firing patterns of neurons in animal nervous systems. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1967
- Accession Number
- AD0658886
Entities
People
- Jack D. Cowan
Organizations
- University of Chicago