Oleksandr V. Popovych
Self-organized noise resistance of oscillatory neural networks with spike timing-dependent plasticity
Oleksandr V. Popovych1, Serhiy Yanchuk2, Peter A. Tass1,3
1 Institute of Neuroscience and Medicine – Neuromodulation (INM-7), Research Center Jülich, 52425 Jülich, Germany
2 Institute of Mathematics, Humboldt University of Berlin, 10099 Berlin, Germany
3 Department of Neuromodulation, Cologne Medical School, 50924 Cologne, Germany
It is known that synchronization in oscillatory networks can be destroyed by independent noise. One might thus expect that such a random noise stimulation can be a powerful tool for desynchronizing synchronized neural populations. We here show that for oscillatory neural networks with adaptive synaptic weights governed by spike timing-dependent plasticity (STDP) intriguingly the opposite is true. In fact, we reveal that independent noise has a constructive effect on the collective dynamics of oscillatory neural ensembles with STDP. According to experimental results, the synaptic coupling strength among neurons is potentiated or depressed depending on whether the firing of the pre-synaptic neuron advances or follows that of the post-synaptic neuron, respectively. We found that the mean synaptic coupling of the neural ensembles increases with the noise intensity, and there is an optimal noise, where the amount of synaptic coupling gets maximal in a resonance-like manner as found for the stochastic or coherence resonances, although the mechanism in our case is different. This leads to a noise-induced self-organization of the coupling topology, which effectively counteracts the desynchronizing impact of independent noise over a wide range of the noise intensity. This phenomenon essentially relies on the presence of STDP. Our results suggest a possible mechanism of how the brain may counteract external perturbations and noise in order to preserve the existing level of neural synchrony and bridge the transition from information coding by precise spiking times to variable and imperfect spike timing. Furthermore, our results show that independent noise can by no means be considered as an effective method for desynchronization of oscillatory neural networks with STDP. Given the attempts to counteract neural synchrony underlying tinnitus with noisers and maskers, our findings may even have clinical relevance and contribute to a deeper understanding of why the maskers and noisers have a limited efficacy in counteracting tinnitus.