Asymmetries in the intrinsic properties of individual LNs and directed connections between neurons can ensure that specific sequences of spiking are stabilized. In addition to changes in the intrinsic

and synaptic parameters, a neuron’s Ca2+ concentration can also influence the ordering of LN bursts (Ahn et al., 2010). The larger the Ca2+ level in a cell, the less likely selleck screening library it is to generate a spike in the next cycle, because of the negative impact of Ca2+-dependent potassium currents. In Figure 1E note that the ordering of bursts in a three-color network can be predicted by the level of Ca2+. Network geometry also influenced other attributes of individual bursts. In a network with chromatic number two (a minimum of two colors was required to color the network), the burst duration of neurons associated with a color depended on the number of neurons in the group. Asymmetries in the network’s structure were manifest as asymmetries in burst duration: larger groups

dominated the dynamics (Figure 2A). The average duration of a burst was a nonlinear function of the number of neurons BMN 673 concentration associated with the group and showed a sharp transition as the group size grew (Figure 2B). For a given network, as the number of separate groups increased, the number of neurons associated with each color decreased and thus exerted diminished inhibitory influence upon other groups. And as this influence diminished, strict coloring-based dynamics tended to break down (Figure 2C, compare top and bottom panels). We quantified this loss of structure as the amount of overlap in activity between different groups (Figure 2D). For a selected network with 100 inhibitory cells, this overlap showed an abrupt transition when the number of colors increased from four to five. We simulated a larger network

consisting of 200 neurons to see if this transition was determined by the size of the network. A similar abrupt increase in variability was also evident in this larger network. Increasing the time constant of recovery from adaptation, however, shifted the transition point to the right (not shown). This transition was also seen in the distribution of burst lengths across all groups (Figure 2E). Linifanib (ABT-869) For a chromatic number below four, the burst length for all groups was very narrowly distributed. When the number of colors exceeded four, the standard deviation of the distribution increased abruptly (Figure 2E). This suggests that below a threshold level of inhibition, neurons showed very low within-group variability. In addition, the simple periodic sequences observed in networks with few colors were replaced by more complex sequences of activity when the number of colors increased. However, even for networks with high chromatic numbers (∼8–10) (bottom panel of Figure 2C), the influence of graph coloring continued to be evident in the network’s dynamics.