DETECTING DYNAMICAL INTERDEPENDENCE AND GENERALIZED SYNCHRONY THROUGHMUTUAL PREDICTION IN A NEURAL ENSEMBLE

A method to characterize dynamical interdependence among nonlinear sys tems is derived based on mutual nonlinear prediction. Systems with non linear correlation will show mutual nonlinear prediction when standard analysis with linear cross correlation might fail. Mutual nonlinear p rediction also provides information on the directionality of the coupl ing between systems. Furthermore, the existence of bidirectional mutua l nonlinear prediction in unidirectionally coupled systems implies gen eralized synchrony. Numerical examples studied include three classes o f unidirectionally coupled systems: systems with identical parameters, nonidentical parameters, and stochastic driving of a nonlinear system . This technique is then applied to the activity of motoneurons within a spinal cord motoneuron pool. The interrelationships examined includ e single neuron unit firing, the total number of neurons discharging a t onetime as measured by the integrated monosynaptic reflex, and intra cellular measurements of integrated excitatory postsynaptic potentials (EPSP's). Dynamical interdependence, perhaps generalized synchrony, w as identified in this neuronal network between simultaneous single uni t firings, between units and the population and between units and intr acellular EPSP's.