Fig. 2: Topographically connected networks produce structured fluctuations without altering neuronal spiking dynamics.

a Schematic diagram of a 2D spiking network model with 80% excitatory (gray) and 20% inhibitory (blue) neurons wired with a uniformly random connection probability. b Spike rasters from 10,000 excitatory neurons along a 1D slice arranged by linear distance in the network. LFP fluctuations calculated from summed synaptic currents for a single 10 × 10 neuron pool is plotted in red. The mean spike rate within one neuron pool is shown in black. c Spatial organization of LFP amplitude for each neuron pool in the network plotted at one time point. d Network schematic as in (a), but the network was topographically connected with probabilities drawn from a Gaussian (σ = 400 µm), and activity had a distance-dependent transmission delay (0.2 m/s). e Spike rasters as in (b), but sparse structured fluctuations were apparent across the network. f Spatial LFP amplitude as in (c), but the LFP was heterogeneous across the network with topographic structure. g The distribution of single-unit mean firing rates did not differ between the random (blue line) and topographic networks (purple line; N = 5000 neurons; p = 0.28, two-tailed Wilcoxon’s rank-sum test). h The distribution of single-unit CV did not differ between the random and topographic networks (p = 0.11). i Pairwise spike coherence did not differ between the random (blue line) and topographic networks (purple line; N = 10 paired adjacent neuron pools; CI test, α = 0.05; dotted lines 95% CI). j Power spectral density for LFP from the random and topographically connected networks.