Figure 1
From: Slow points and adiabatic fixed points in recurrent neural networks
Phase diagram of the three-neuron system showing five different phases distinguished by the stability and the shape of the adiabatic fixed points (AFP) for different values of the adaption rate ε. One can observe three second order transitions (dashed) and one first order transition showing hysteresis (striped area). The subplots show the firing rate (green), saddle node AFP (gray), stable AFP (blue) and effectively attracting AFP (red) over time and represent a phase each.