RÉSUMÉ
A time series from a human electroencephalogram (EEG) is used as a local perturbation to a reaction-diffusion model with spatiotemporal chaos. For certain finite ranges of amplitude and frequency it is observed that the strongly irregular perturbations can induce transient coherence in the chaotic system. This could be interpreted as "on-line" detection of an inherently correlated pattern embedded in the EEG.
Sujet(s)
Encéphale/physiologie , Électroencéphalographie , Dynamique non linéaire , Humains , Modèles biologiquesRÉSUMÉ
We introduce a nonlinear dynamical system with self-exciting chaotic dynamics. Its interspike interval return map shows a noisy Poisson-like distribution. Spike sequences from different initial conditions are unrelated but possess the same mean frequency. In the presence of noisy perturbations, sequences started from different initial conditions synchronize. The features of the model are compared with experimental results for irregular spike sequences in neurons. Self-exciting chaos offers a mechanism for temporal coding of complex input signals.