Abstract
In the brain, synchronization among cells of an assembly is a common phenomenon and is thought to be functionally relevant. Here, we used an in vitro experimental model of cell assemblies—cortical cultures—combined with numerical simulations of a spiking neural network (SNN) to investigate how and why spontaneous synchronization occurs. To focus on excitatory dynamics, we pharmacologically blocked GABAAergic transmission using bicuculline. Synchronous events in cortical cultures tend to involve almost every cell and display relatively constant durations, which we named "network spikes" (NS). The inter-NS-intervals (INSIs) proved to be a more interesting phenomenon. In most cortical cultures, NSs typically come in series or bursts ("bursts of NSs", BNS), with short (~1 s) INSIs separated by long silent intervals (tens of seconds), leading to bimodal INSI distributions. This suggests that a facilitating mechanism is at work, presumably short-term synaptic facilitation, alongside two fatigue mechanisms: one with a short timescale (short-term synaptic depression) and another with a longer timescale (cellular adaptation). We incorporated these three mechanisms into the SNN, which successfully produced realistic BNSs. By systematically varying recurrent excitation, we found that experimental cultures operate within an intermediate weakly-synchronized regime (CV~0.5), with an adaptation timescale in the 2-8 s range. Our results demonstrate that INSI statistics allow us to infer hidden biological parameters that are otherwise inaccessible experimentally.
Introduction
In both living brains and lab-grown cultures, neurons do not fire in isolation; they often synchronize in massive, brief bursts known as "network spikes" (NS). While these events are well-documented, the underlying mechanisms governing their timing—specifically why they often cluster into "bursts of network spikes" (BNS) separated by long silences—remain poorly understood. Previous computational models often failed to replicate this complex timing because they typically relied on a single fatigue mechanism. This study addresses the gap by proposing a unified model that explains the bimodal distribution of inter-event intervals through a combination of three distinct adaptive mechanisms: facilitation, depression, and cellular adaptation.
Methods
The study employed a dual approach, integrating biological experiments with computational modeling.
Experimental Model
Rat cortical neurons were grown on multielectrode arrays (MEAs). To isolate excitatory dynamics, inhibitory (GABAergic) transmission was pharmacologically blocked using bicuculline. This simplified system allowed researchers to focus on how recurrent excitation alone drives synchronization.
Computational Model
A Spiking Neural Network (SNN) was constructed using 800 Leaky Integrate-and-Fire (LIF) neurons. The model's core innovation was the integration of three adaptive mechanisms:
- Short-Term Facilitation (STF): A positive feedback mechanism where recent activity strengthens synapses.
- Short-Term Depression (STD): A fast fatigue mechanism where synapses weaken temporarily after firing.
- Cellular Adaptation: A slow fatigue mechanism (after-hyperpolarization current) that increases the threshold for neuronal excitation over longer periods.
Results
The research revealed that all three mechanisms are essential to replicate realistic bursting behavior. STD is responsible for terminating individual network spikes, STF promotes the rapid succession of spikes within a burst, and cellular adaptation enforces the long recovery periods between bursts.
The statistical distribution of Inter-Network-Spike-Intervals (INSIs) served as a "microscope" into the system's dynamics. By comparing simulated data with biological recordings, the authors identified that real cultures operate in a "weakly-synchronized" regime (Coefficient of Variation ≈ 0.5). This comparison allowed for the inference of hidden parameters, such as a cellular adaptation timescale between 2 and 8 seconds. Furthermore, the generation of BNS was found to follow a "Poisson-with-refractory-period" model, where the refractory period is approximately four times the adaptation timescale.
Discussion
This work demonstrates that the complex rhythms of neuronal assemblies can be decoded using relatively simple statistical measures of their activity. By identifying the mandatory "trio" of adaptive mechanisms, the researchers have provided a framework for understanding how excitatory networks balance promotion and fatigue.
A primary limitation of this study is the exclusion of inhibitory neurons. In a natural brain environment, the balance between excitation and inhibition would likely alter these dynamics. Future work should explore the impact of inhibitory circuits, structural modularity, and Spike-Timing-Dependent Plasticity (STDP) on these network rhythms.
Summary updated