Abstract
It has previously been shown that the encoding of time-dependent signals by feedforward networks (FFNs) of processing units exhibits suprathreshold stochastic resonance (SSR), which is an optimal signal transmission for a finite level of independent, individual stochasticity in the single units. In this study, a recurrent spiking network is simulated to demonstrate that SSR can be also caused by network noise in place of intrinsic noise. The level of autonomously generated fluctuations in the network can be controlled by the strength of synapses, and hence the coding fraction (our measure of information transmission) exhibits a maximum as a function of the synaptic coupling strength. The presence of a coding peak at an optimal coupling strength is robust over a wide range of individual, network, and signal parameters, although the optimal strength and peak magnitude depend on the parameter being varied. We also perform control experiments with an FFN illustrating that the optimized coding fraction is due to the change in noise level and not from other effects entailed when changing the coupling strength. These results also indicate that the non-white (temporally correlated) network noise in general provides an extra boost to encoding performance compared to the FFN driven by intrinsic white noise fluctuations.
Highlights
The role of neurons as signal encoders has been shown to be enhanced in some situations by noise through the phenomenon known as stochastic resonance (SR), in which a finite amount of noise linearizes the response of a single processing unit or a population to a weak signal (Gammaitoni et al, 1998; Lindner et al, 2004; McDonnell & Ward, 2011)
The question that was being investigated was whether the benefits to signal encoding, especially suprathreshold stochastic resonance (SSR), awarded by the increase in intrinsic white noise and the heterogeneity of parameters for feedforward networks (Stocks, 2000; Stocks & Mannella, 2001; Hoch et al, 2003; Das et al, 2009; Ashida & Kubo, 2010; Nikitin et al, 2010; Durrant et al, 2011; Hunsberger et al, 2014; Beiran et al, 2017) could be found in recurrent networks whose synaptic weights were increased instead
SSR was shown to occur with respect to the level of network noise controlled by the synaptic coupling strength and the effect was robust over a wide range of intrinsic, network, and signal parameters
Summary
The role of neurons as signal encoders has been shown to be enhanced in some situations by noise through the phenomenon known as stochastic resonance (SR), in which a finite amount of noise linearizes the response of a single processing unit or a population to a weak signal (Gammaitoni et al, 1998; Lindner et al, 2004; McDonnell & Ward, 2011). We test whether the observed encoding could be obtained in control experiments in which the network input is replaced by a constant input, adjusted such that neurons have the same firing rate as neurons in the recurrent network (RN), or by a constant input and an individual Gaussian noise, adjusted according to the diffusion approximation. In the latter case does one observe an enhanced encoding. This demonstrates that the optimal coding fraction at a nonzero synaptic amplitude is due to the network noise and not due to other effects that a variation of coupling strength might entail
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have