Emergent bimodal firing patterns implement different encoding strategies during gamma-band oscillations

Abstract

Upon sensory stimulation, primary cortical areas readily engage in narrow-band rhythmic activity between 30 and 90 Hz, the so-called gamma oscillations. Here we show that, when embedded in a balanced network, type-I excitable neurons entrained to the collective rhythm show a discontinuity in their firing-rates between a slow and a fast spiking mode. This jump in the spiking frequencies is characteristic to type II neurons, but is not present in the frequency-current curve (f-I curve) of isolated type I neurons. Therefore, this rate bimodality arises as an emerging network property in type I population models. We have studied the mechanisms underlying the generation of these two firing modes, in order to reproduce the spiking activity of in vivo cortical recordings, which is known to be highly irregular and sparse. We have also analyzed the relation between afferent inputs and the single unit activity, and between the latter and the local field potential (LFP) phase, in order to establish how the collective dynamics modulates the spiking activity of the individual neurons. Our results reveal that the inhibitory-excitatory balance allows two encoding mechanisms, for input rate variations and LFP phase, to coexist within the network.

Temporally correlated fluctuations drive epileptiform dynamics

“…Recent studies of OU processes driving neural models have investigated the effects of coloured noise on temporal distributions of neuronal spiking (Braun et al, 2015, da Silva and Vilela, 2015) and the generation of multimodal patterns of alpha activity (Freyer et al, 2011). In addition, networks of spiking neurons (Sancristóbal et al, 2013) and of neuronal populations (Jedynak et al, 2015) have been shown to generate realistic $1/f^b\mathrm{-}\mathrm{like}$ spectra when driven by OU noise, or more complex dynamics when subjected to driving at specific frequencies (Spiegler et al, 2011, Malagarriga et al, 2015). However, we lack an understanding of the ways in which non-white noise or rhythmic perturbations interact with neuronal populations to produce epileptiform dynamics.…”

Cross-frequency transfer in a stochastically driven mesoscopic neuronal model

“…The correlation time τ dictated by synaptic effects is conjectured to be of the order of 10 ms (Mazzoni et al, 2008 ; Sancristóbal et al, 2013 ) rather than 100 ms. In our case, however, noise stands for background activity arising from collective effects at the mesoscopic scale.…”

“…This is a reasonable assumption, since the brain has multiple sources of noise (Faisal et al, 2008 ) that have a variety of functional roles (McDonnell and Ward, 2011 ). In microscopic models, a temporally correlated Ornstein-Uhlenbeck noise is known to reproduce the observed 1/ f spectral profile of LFP activity (Sancristóbal et al, 2013 ). Our results show that a network of coupled neural masses subject to temporally correlated noise exhibits a well-defined rhythm (in the alpha range) embedded in a broadband spectral background similar to what is observed experimentally.…”

Probing scale interaction in brain dynamics through synchronization

“…The rate functions a x and b x for each gating variable, together with all the NN parameters used throughout this paper are given in [10]. The synaptic current I syn is described using a conductance-based formalism…”

“…This allows the brain to be traditionally investigated in a reductionist way, using different simplified levels of description. This approach has been very fruitful in unveiling several mechanisms that lay at the basis of the observed neural tissue behaviour [7][8][9][10].…”