Suprathreshold stochastic resonance in neural processing tuned by correlation

Abstract

Suprathreshold stochastic resonance (SSR) is examined in the context of integrate-and-fire neurons, with an emphasis on the role of correlation in the neuronal firing. We employed a model based on a network of spiking neurons which received synaptic inputs modeled by Poisson processes stimulated by a stepped input signal. The smoothed ensemble firing rate provided an output signal, and the mutual information between this signal and the input was calculated for networks with different noise levels and different numbers of neurons. It was found that an SSR effect was present in this context. We then examined a more biophysically plausible scenario where the noise was not controlled directly, but instead was tuned by the correlation between the inputs. The SSR effect remained present in this scenario with nonzero noise providing improved information transmission, and it was found that negative correlation between the inputs was optimal. Finally, an examination of SSR in the context of this model revealed its connection with more traditional stochastic resonance and showed a trade-off between supratheshold and subthreshold components. We discuss these results in the context of existing empirical evidence concerning correlations in neuronal firing.

RETRACTED: Local Correlated Noise Improvement of Signal-to-Noise Ratio Gain in an Ensemble of Noisy Neuron

“…Recently, the enhancement of SNR gain (larger than unity) is reliably observed in parallel arrays of nonlinear subsystems assisted by the independent internal noises [11] [12] [13] [14]. This regular model of uncoupled parallel arrays of nonlinear subsystems elicits many important mechanisms of non-conventional SR effects, such as, SR without tuning [7], suprathreshold SR [9] and array SR [11]. In such an ensemble, all units have a common input, and their outputs are summed as the collective response [11] [12] [13] [14].…”

Decoding suprathreshold stochastic resonance with optimal weights

“…Moreover, artificial sensors, digital beamforming, biological neurons, cochlear implants and multiaccess communication systems can all be unified under the concept of stochastic pooling networks that manifest the noiseenhanced processing property [32][33][34]. Due to the variety of scenarios where SSR is observed, a number of performance measures have been considered, for instance, mutual information [22,23,27,[29][30][31], mean square error (MSE) distortion [35,41,43], inputoutput cross-correlation [35,38], Fisher information [36,39,43] and signal-to-noise ratio [37].…”

A review of methods for identifying stochastic resonance in simulations of single neuron models

“…Suprathreshold stochastic resonance is a form of stochastic facilitation that can occur in populations of neurons organized in parallel (Stocks and Mannella 2001;McDonnell et al 2008;McDonnell and Stocks 2009;Ashida and Kubo 2010;Durrant et al 2011). It is unlike classical stochastic resonance in several ways, the most important of which is that noise enables a more precise compressed encoding to be made of an analog variable than is possible without noise; it is likely to be most relevant to the sensory periphery (McDonnell et al 2008).…”