The behavior of a neural network model for binocular rivalry is explored through the development of an analogy between it and an electronic astable multivibrator circuit. The model incorporates reciprocal feedback inhibition between signals from the left and the right eyes prior to binocular convergence. The strength of inhibitory coupling determines whether the system undergoes rivalrous oscillations or remains in stable fusion: strong coupling leads to oscillations, weak coupling to fusion. This implies that correlation between spatial patterns presented to the two eyes can affect the strength of binocular inhibition. Finally, computer simulations are presented which show that a reciprocal inhibition model can reproduce the stochastic behavior of rivalry. The model described is a counterexample to claims that reciprocal inhibition models as a class cannot exhibit many of the experimentally observed properties of rivalry.