INFLUENCE OF SYNAPTIC NOISE AND DYNAMICAL SYNAPSES ON NETWORK ACTIVITY AND SYNCHRONY

Synapses in the brain are not fixed. Unreliable neurotransmitter release and noise in synaptic receptor currents make synapses inherently random. Here, we investigated the consequences of multiplicative noise in synaptic weights on the dynamics of a recurrent network of leaky-integrate-and-fire neurons. We found that synaptic noise increased the variance of firing rates and synchronized the network. This was due to the increase in shot noise events. Analysis of synaptic currents revealed that noise broadened the distribution of synaptic currents in the network, which synchronised some neurons and silenced others. While the transfer function of the network depended more strongly on the mean synaptic weight, the introduction of noise made it more dispersed resulting in a wide range of firing rates.

Next, we show that besides stochasticity in neurotransmitter release, short-term plasticity is a strong source of multiplicative noise when neurons are firing irregularly. We found that synaptic facilitation increased the mean and variance of synaptic currents leading to heavy tailed distribution of firing rates while depression led to a narrow distribution of synaptic currents and firing rates. Varying the parameters of short-term plasticity regulated the amount of synaptic noise and affected both the dynamic and noise regime of the network by changing both the linearity and dispersion of the transfer function. Stronger synaptic depression led to a linear and input-driven regime, while stronger synaptic facilitation resulted in nonlinearity and synchrony.

Thus, our results shed light on the general consequences of temporal fluctuations in synaptic weights in a biological neural network.