ABSTRACT
ABSTRACT We have studied some of the functional aspects of reciprocally connected neural elements using both computer simulations and in vivo studies. Computer simulations were conducted using the general purpose neural simulation package GENESIS. Two biologically realistic network models were constructed; the first consisted of a pair of interconnected excitatory and inhibitory neurons, and the second model extended the first to include long-range corticocortical connections. Both of these models exhibited chaotic dynamics that depended on the strength of the reciprocal connections. The behavior of these networks were explored using the methods of bifurcation diagrams, phase plots, and Fano-factor analysis, which revealed the fractal nature of the simulated spike trains. The chaotic dynamics of the real brain were studied by recording field potentials from the cortex of chronically implanted rats during anesthetized and awake states. The correlation dimension of the field potentials was found to decrease under the effects of anesthesia, possibly because of decreases in the strength of reciprocal connections.
