Simulation of oscillatory activity in the gamma frequency range in a network of interneurons coupled by fast inhibitory synapses. Aa, Ab, Simulated voltage traces of 15 neurons and rastergram representation of the activity of a network of 100 neurons connected randomly by rapid synapses (synaptic delay 0.8 msec; τ_d = 1.8 msec;g_Syn = 0.02 mS/cm²;M_Syn = 60). The excitatory driving current was heterogeneously distributed withI_μ = 3 μA/cm²and I_ς = 0.09 μA/cm². Ac, Rastergram of the activity of a network coupled by slow inhibitory synapses (τ_d = 5.2 msec), as observed at the BC–GC synapse. Note the lower network frequency. Also note that the activity of 13 neurons is suppressed. B, C, Mean network frequency (f_μ, B) and coherence (κ, C), plotted against the mean current drive (I_μ,I_ς/I_μ = 0.03) for fast (τ_d = 1.8 msec, ●) and slow decay time constant of the inhibitory postsynaptic conductance change (τ_d = 5.2 msec, ○). Arrows indicate data points corresponding to the rastergrams shown in Ab and Ac. D, Coherence κ plotted against the heterogeneity measureI_ς/I_μ of the current drive for I_μ = 1, 2, 3, and 5 μA/cm² (▪, ●, ▴, and ♦, respectively).E, Plot of network coherence κ versusM_Syn and I_μ. Coherence κ increases with both M_Syn andI_μ. Note that critical values of the two parameters for the generation of coherent oscillations are interdependent.