COUPLED CIRCLE MAPS AS A TOOL TO MODEL SYNCHRONIZATION IN NEURAL NETWORKS

Temporal correlation and decorrelation of the spiking of groups of neurons have been suggested to be of importance for the segmentation of different features to objects (binding problem). We show that coupled circle maps exhibiting chaotic oscillations are a useful tool to simulate the behaviour of such systems. In a model where one map represents the phase dynamics of one neuron or a group of neurons we observe that, depending on the coupling strength, the different maps show correlated or uncorrelated behaviour, while the autocorrelation function remains flat, as expected for a chaotic signal. This synchronized behaviour can be organized by a simple Hebb-type learning rule.