Figure 2

Stochastic spiking network model reproduces tectal bursting.

(A) LNP Model architecture: connection weights for excitatory (E) and inhibitory (I) interactions are determined by Gaussian functions of the intercellular Euclidian distance, with unique gain (g^(E),g^(I)) and spatial standard deviation (σ^(E),σ^(I))for each type of interaction. Presynaptic spikes are filtered in time with time constants (τ^(E),τ^(I))and summed along with a bias (μ) to produce the linear drive (excitability state). Exponentiating this linear drive sets the mean of an inhomogeneous Poisson process from which spikes are randomly emitted (Materials and methods). Model parameters used for all panels in this figure are shown inset. (B) Example of burst detection in the simulation results (c.f. Figure 1F)(C) Left: Locations of bursts detected during a 30 min simulation. Colors indicate burst times and spot sizes indicate number of participating neurons. Right: Burst locations vs. burst time. Rectangle widths indicate the temporal extent of each burst. (D–E) Neuronal assemblies detected using PCA-promax algorithm (Materials and methods), for simulation results (D) and experimental data (E), shown on a spatial map (top) and a raster plot for representative assemblies (bottom). (F) Time-course of linear drive to model cells around the time of a spontaneous burst (t=0). Star indicates burst centroid.