Difference in coherence levels and frequency tuning of young and mature interneuron network models. A, Raster plot illustrates the activity of the interneuron network model with mature properties at the maximal coherence level. Each dot represents an action potential during one simulation run (−150 to 500 ms); index of neurons (1–200) is plotted on the y-axis. The graph below the raster plot represents the spike-time histogram (bin width, 1 ms; maximal amplitude, 100 neurons). Tonic excitation was applied at random time points (−150 ms ≤ t < −100 ms); inhibitory synapses were enabled at t ≥ 0 ms. Coherence was determined during the last 100 ms of the simulation period. B–I, Three-dimensional plots of the coherence measure κ versus the tonic excitatory drive (I_μ) and the unitary peak conductance (g_syn). The height of the peaks in the three-dimensional plots indicates the level of synchrony in the network, quantified by κ. Corresponding network frequencies (f_μ) are indicated by the color code superimposed on the surface of the plots (see scale bar at the bottom). The heterogeneity of the excitatory drive (I_{σ /}I_μ) was 10%; the reversal potential of the inhibitory conductance was −55 mV, reflecting shunting inhibition between P6 and P25 (Chavas and Marty, 2003; Vida et al., 2006; Banke and McBain, 2007). For additional details, see Materials and Methods. Simulations were started with a mature network (B). Subsequently, intrinsic, synaptic, and network parameters were changed stepwise to reproduce properties of the young network (I). The mature interneuron network (B) generates gamma oscillations with high coherence (maximal κ = 0.74) in the upper gamma frequency range (50–90 Hz; orange). Introduction of young intrinsic properties (prop.), including slow intrinsic membrane properties, broad action potentials, and slow action potential conduction velocity (C) or slow decay of the IPSC (D), decreases coherence and reduces network frequency to the lower gamma range (30–50 Hz; red). Introduction of low network connectivity (E; M_syn = 40) also reduces coherence, but it keeps oscillatory frequency largely unchanged (50–90 Hz; yellow). Surprisingly, the combination of young intrinsic properties with slow IPSCs (F) results in an increase in peak coherence, whereas network frequency remains low. In contrast, the combination of low connectivity with young intrinsic properties (G) or slow IPSCs (H) results in a further marked reduction in coherence. When all properties are combined to simulate the young network (I), oscillations are generated at low coherence (maximal κ = 0.39) in the lower gamma range (30–50 Hz). J, Raster plot of network activity in a network with young properties at the maximal level of coherence. The asterisk and triangle in B and I indicate parameter settings for the raster plots.