RECONSTRUCTING CHAOTIC DYNAMICS THROUGH SPIKE FILTERS

We consider the problem of reconstructing chaotic attractors from spike trains produced by model neurons. We find that the period-parameter plot, which displays the dependence of the oscillation frequency of the spiking model on the input level, is a useful device for determining the success of reconstruction. A three-dimensional version of Fitzhugh-Nagumo spiking dynamics is investigated along with other models.