Dynamics of Image Formation by Nerve Cell Assemblies

Abstract

Masses of nerve cells that comprise the cerebral cortex have certain properties that promote the emergence of cooperative ensemble. These include the immense numbers of quasi-autonomous elements, the neurons, the high density of interconnections by axons and dendrites at synapses, the linear operations of temporal and spatial integration, and the regenerative nonlinearities of neural function. The ensemble differs from the “average neuron” in the way ice differs from an “average molecule” of water. The time and distance scales of macroscopic neural mass action are one to three orders of magnitude greater than the scales of microscopic neural action. The state variables are activity density functions distributed in both time and space, and the operations on the state variables differ from their homologs for single neurons. Analysis of the dynamics of a primitive form of cortex show that sensory information is encoded in macroscopic patterns of neural activity analogous to the way that temperature reflects kinetic energy of particles in a fluid. A key step of information processing depends on an orderly transition of a cortical neural mass from a quasi-equilibrium state to a limit cycle state and back again. The limit cycle activity serves as an operator on sensory input to abstract and generalize aspects of the input into pre-established categories, thereby creating information for further central processing.