Noise-driven switching between limit cycles and adaptability in a small-dimensional excitable network with balanced coupling

We study a system of globally coupled FitzHugh-Nagumo equations. Each unit is either excitatory or inhibitory. If the numbers of units of both types are in a specific ratio, we observe the presence of multistable oscillatory states with different excitation or firing rates. In the presence of noise, there is noise-driven switching between these states and the resultant firing pattern is long-range correlated. The choice between higher and lower frequency oscillations depends on the input, which results in increasing adaptability of the system's output to the periodic input