Emergent population activity in metric-free and metric networks of neurons with stochastic spontaneous spikes and dynamic synapses

We show that networks of excitatory neurons with stochastic spontaneous spiking activity and short-term synaptic plasticity can exhibit spontaneous repetitive synchronization in so-called population spikes. The major reason for this is that synaptic plasticity nonlinearly modulates the interaction between neurons. For large-scale two-dimensional networks, where the connection probability decreases exponentially with increasing distance between the neurons resulting in a small-world network connectome, a population spike occurs in the form of circular traveling waves diverging from seemingly non-stationary nucleation sites. The latter is in drastic contrast to the case of networks with a fixed fraction of steady pacemaker neurons, where the set of a few spontaneously formed nucleation sites is stationary. Despite the spatial non-stationarity of their nucleation, population spikes may occur surprisingly regularly. From a theoretical viewpoint, these findings show that the regime of nearly-periodic population spikes, which mimics respiratory rhythm, can occur strictly without stochastic resonance. In addition, the observed spatiotemporal effects serve as an example of transient chimera patterns.

Temporal discrimination from the interaction between dynamic synapses and intrinsic subthreshold oscillations

The interaction between synaptic and intrinsic dynamics can efficiently shape neuronal input-output relationships in response to temporally structured spike trains. We use a neuron model with subthreshold oscillations receiving inputs through a synapse with short-term depression and facilitation to show that the combination of intrinsic subthreshold and synaptic dynamics leads to channel-specific nontrivial responses and recognition of specific temporal structures. We employ the Generalized Integrate-and-Fire (GIF) model, which can be subjected to analytical characterization. We map the temporal structure of spike input trains to the type of spike response, and show how the emergence of nontrivial input-output preferences is modulated by intrinsic and synaptic parameters in a synergistic manner. We demonstrate that these temporal input discrimination properties are robust to noise and to variations in synaptic strength, suggesting that they likely contribute to neuronal computation in biological circuits. Furthermore, we also illustrate the presence of these input-output relationships in conductance-based models.Author summaryNeuronal subthreshold oscillations underlie key aspects of information processing in single neuron and network dynamics. Dynamic synapses provide a channel-specific temporal modulation of input information. We combine a neuron model that displays subthreshold oscillations and a dynamic synapse to analytically assess their interplay in processing trains of spike-mediated synaptic currents. Our results show that the co-action of intrinsic and synaptic dynamics builds nontrivial input-output relationships, which are resistant to noise and to changes in synaptic strength. The discrimination of a precise temporal structure of the input signal is shaped as a function of the joint interaction of intrinsic oscillations and synaptic dynamics. This interaction can result in channel-specific recognition of precise temporal patterns, hence greatly expanding the flexibility and complexity in information processing achievable by individual neurons with respect to temporal discrimination mechanisms based on intrinsic neuronal dynamics alone.