Sample simulations of a large spiking network in the III scenario. Three different samples of spiking network simulations are shown, corresponding to the three different dynamical behaviors of the III network with 24000 neurons. (A) \(a=0.75\), \(b=0.75\), where all the populations are active. (B) \(a=1.4\), \(b=1.0\), where the subnetworks oscillate with a phase difference of 2π/3. (C) \(a=2.0\), \(b=2.0\), where only one inhibitory population is active, persistently dominating the other two populations. (D–F) Population firing rates corresponding to the neural spiking dynamics in (A–C), respectively, obtained by counting the spikes generated by each subnetwork in time bins of width 3 ms. (G) State space embedding of the firing rates of the three subnetworks for \(a=0.75\), \(b=0.75\). The activity fluctuates around a fixed point that represents a stable equilibrium. (H) State space illustration of the firing rates of the three subnetworks for \(a=1.4\), \(b=1.0\). In this case, oscillations of increasing period emerge, and the state space representation is a closed trajectory, after a long enough time has elapsed. (I) State space illustration of the firing rates of the three subnetworks for \(a=2.0\), \(b=2.0\), where the trajectory converges to a single attractive fixed point and remains there.