Neuron models with a continuous gain function are called type I. Mathematically, a saddle-node-onto-limit-cycle bifurcation generically gives rise to a type I behavior, as we will explain now.

For zero input and weakly positive input, we suppose that our neuron model has three fixed points in a configuration such as in Fig. 4.13. A stable fixed point (node) to the left, a saddle point in the middle, and an unstable fixed point to the right. If $I$ is increased, the $u$-nullcline moves upward and the stable fixed point merges with the saddle and disappears (Fig. 4.12). We are left with the unstable fixed point around which there must be a limit cycle provided the flux is bounded. If the limit cycle passes through the region where the saddle and node disappeared, the scenario is called a saddle-node-onto-limit-cycle bifurcation.

Can we say anything about the frequency of the limit cycle? Just before the transition point where the two fixed points merge, the system exhibits a stable fixed point which (locally) attracts trajectories. As a trajectory gets close to the stable fixed point, its velocity decreases until it finally stops at the fixed point. Let us now consider the situation where the driving current is a bit larger so that $I=I_{\theta}$. This is the transition point, where the two fixed points merge and a new limit cycle appears. At the transition point the limit cycle has zero frequency because it passes through the two merging fixed points where the velocity of the trajectory is zero. If $I$ is increased a little, the limit cycle still ‘feels’ the ’ghost’ of the disappeared fixed points in the sense that the velocity of the trajectory in that region is very low. While the fixed points have disappeared, the ‘ruins’ of the fixed points are still present in the phase plane. Thus the onset of oscillation is continuous and occurs with zero frequency. Models which fall into this class are therefore of type I; cf. Fig. 4.11.

From the above discussion it should be clear that, if we increase $I$, we encounter a transition point where two fixed points disappear, viz., the saddle and the stable fixed point (node). At the same time a limit cycle appears. If we come from the other side, we have first a limit cycle which disappears at the moment when the saddle-node pair shows up. The transition is therefore called a saddle-node bifurcation on a limit cycle.

There is no fundamental reason why a limit cycle should appear at a saddle-node bifurcation. Indeed, in one-dimensional differential equations, saddle-node bifurcations are possible, but never lead to a limit cycle. Moreover, if a limit cycle exists in a two-dimensional system, there is no reason why it should appear directly at the bifurcation point - it can also exist before the bifurcation point is reached. In this case, the limit cycle does not pass through the ruins of the fixed point and therefore has finite frequency. This gives rise to a type II neuron model. The corresponding bifurcation can be classified as Saddle-Node-off-Limit-Cycle.

The typical jump from zero to a finite firing frequency, observed in the frequency-current curve of type II models can arise by different bifurcation types. One important example is a Hopf bifurcation.

Let us recall that the fixed points of the system lie at the intersection of the $u$-nullcline with the $w$-nullcline. In the FitzHugh-Nagumo model, with parameters as in Fig. 4.10, there is always a single fixed point whatever the (constant) driving current $I$. Nevertheless, while $I$ is slowly increased, the behavior of the system changes qualitatively from a stable fixed point to a limit cycle; cf. Fig. 4.10. The transition occurs when the fixed point loses its stability.

From the solution of the stability problem in Eq. (4.25) we know that the Eigenvalues $\lambda_{+/-}$ form a complex conjugate pair with a real part $\gamma$ and a imaginary part $+/-\omega$ (Fig. 4.16). The fixed point is stable if $\gamma<0$. At the transition point, the real part vanishes and the eigenvalues are

These eigenvalues correspond to an oscillatory solution (of the linearized equation) with a frequency given by $\omega=\sqrt{F_{u}G_{w}-G_{u}F_{w}}$. The above scenario of stability loss in combination with an emerging oscillation is called a Hopf-bifurcation.

Unfortunately, the discussion so far does not tell us anything about the stability of the oscillatory solution. If the new oscillatory solution, which appears at the Hopf bifurcation, is itself unstable (which is more difficult to show), the scenario is called a subcritical Hopf-bifurcation (Fig. 4.16). This is the case in the FitzHugh-Nagumo model where, due to the instability of the oscillatory solution in the neighborhood of the Hopf bifurcation, the system blows up and approaches another limit cycle of large amplitude; cf. Fig. 4.10. The stable large-amplitude limit cycle solution exists, in fact, slightly before $I$ reaches the critical value of the Hopf bifurcation. Thus there is a small regime of bistability between the fixed point and the limit cycle.

In a supercritical Hopf bifurcation, on the other hand, the new periodic solution is stable. In this case, the limit cycle would have a small amplitude if $I$ is just above the bifurcation point. The amplitude of the oscillation grows with the stimulation $I$ (Fig. 4.16). Such periodic oscillations of small amplitude are not linked to neuronal firing, but must rather be interpreted as spontaneous subthreshold oscillations.

Whenever we have a Hopf bifurcation, be it subcritical or supercritical, the limit cycle starts with finite frequency. Thus if we plot the frequency of the oscillation in the limit cycle as a function of the (constant) input $I$, we find a discontinuity at the bifurcation point. However, only models with a subcritical Hopf-bifurcation give rise to large-amplitude oscillations close to the bifurcation point. We conclude that models where the onset of oscillations occurs via a subcritical Hopf bifurcation exhibit a gain function of type II.