Zeberg H , Blomberg C , Arhem P .

	Figure 1. Type 1 and Type 2 dynamics in the hippocampal neuron model.The time-course of the membrane voltage with increasing steady current for low and high K channel densities. (A)  = 20 µm/s and  = 5 µm/s. The onset frequency is infinitely small. (B)  = 20 µm/s and  = 5 µm/s. The onset frequency is 30 Hz. Note the damped oscillation with stimulation at 114 mA/m2.
	Figure 2. Oscillation maps for the hippocampal neuron model.(A) Regions in the − plane associated with different threshold dynamics. Oscillations occur within the area defined by the continuous line. Double-limit cycle bifurcations in the A2 region, Andronov-Hopf bifurcations (together with double-limit cycle bifurcations) in the B region and saddle-node bifurcations in the C1 region. The bold dashed line indicates the border for channel densities associated with three stationary potentials. The map is a projection of a curved plane in the −−Istim space (on which the oscillation starts) to the − plane. (B) The corresponding three-dimensional map, showing the volume associated with oscillations in the −−Istim space. Oscillations occur in the volume defined by blue and green surfaces. The green surface area represents double-limit cycle bifurcations and the blue area saddle-node bifurcations (SNICs).
	Figure 3. Bifurcation diagrams for the hippocampal neuron model.(A) A saddle node bifurcation in region C1. There are three stationary voltages in the Istim range of −40 to +50 mA/m2. The oscillations occur when the stable stationary potential Vs1 merges with a saddle point voltage Vs2. Type 1 threshold dynamics is generated if the limit cycle involves the merged point, i.e. a saddle-node bifurcation on an invariant circle (SNIC).  = 20 µm/s,  = 2 µm/s. (B) Subcritical Andronov-Hopf and double-limit cycle bifurcations in region B,  = 20 µm/s,  = 10 µm/s. The oscillations emerge at Istim = 84 mA/m2, thus when the corresponding stationary point/voltage still is stable. The loss of stability is due to a double-limit cycle bifurcation, characterized in the variable space by the simultaneous appearance of two limit cycles of opposite stability, one yielding stable and persistent oscillations. This bifurcation is not detectable by the Jacobian matrix of the stationary point; instead the bifurcation depends on the global properties of the variable space. The local Andronov-Hopf bifurcation (also named degenerate Andronov-Hopf bifurcation because of the way the limit cycles collapse onto the equilibrium point [21], [29]) occurs at Istim = 92 mA/m2. There is also an additional Andronov-Hopf bifurcation at higher Istim (524 mA/m2, now shown) that terminates the oscillations. (C) For higher values of (region A2) these two Andronov-Hopf points collide and disappear (the non-transversal Andronov-Hopf bifurcation), after which no Andronov-Hopf points are present  = 20 µm/s,  = 20 µm/s.
	Figure 4. Prerequisites for three stationary potentials (defining region C1).Steady-state currents versus membrane voltage for the hippocampal neuron model. Calculated from Equation 17. The Na channel density is varied while other parameters are maintained constant to demonstrate the requirement of a high Na channel density to obtain three stationary potentials. Inward currents are shown as positive. (A)  = 30 µm/s and  = 5 µm/s. (B)  = 11 µm/s and  = 5 µm/s.
	Figure 5. Type 2 dynamics within region C1 for the hippocampal neuron model.The time-course of the membrane voltage with increasing steady current.  = 40 µm/s and  = 15 µm/s. The onset frequency is 8 Hz.
	Figure 6. A Andronov-Hopf bifurcation within region C1.Schematic bifurcation diagram showing a subcritical Andronov-Hopf bifurcation within the range of three stationary potentials. The distance between the Andronov-Hopf bifurcation and the coalescence of Vs1 and Vs2 has been extrapolated.
	Figure 7. Revised oscillation maps for the hippocampal neuron model.Regions associated with oscillations in the − plane, showing the existence of Type 2 dynamics within region C1. (A) Onset frequencies. (B) Oscillation map for comparison with the map of Fig. 2, showing the subregions C1a and C1b. The border between C1a and C1b closely follows the Bogdanov-Takens bifurcation curve (see Table 1).
	Figure 8. Bifurcation curves and the three-root solution space for the hippocampal neuron model.Istim− diagrams at  = 40 µm/s. The thick continuous line defines the region associated with three-root solutions of Equation 17. The thin continuous line is the Andronov-Hopf bifurcation curve and the hatched line, defining the oscillation limit, is the double limit cycle bifurcation curve. (A) An overall perspective. (B) A detailed view of the cusp of the three-root solution space to describe the two subregions, defined by the stability of the stationary potentials. The Bogdanov-Takens bifurcation point is marked.
	Figure 9. Exclusively Type 2 dynamics in the myelinated axon model.The time-course of the membrane voltage with increasing steady current for low and high K channel densities. (A)  = 300 µm/s and  = 0 µm/s. The onset frequency is 59Hz. (B)  = 300 µm/s and  = 40 µm/s. The onset frequency is 139Hz.
	Figure 10. Bifurcation curves and the three-root solution space for the myelinated axon model.Istim− diagrams at  = 200 µm/s. The thick continuous line defines the region associated with three-root solutions of Equation 14. The thin continuous line is the Andronov-Hopf bifurcation curve and the hatched line is the double limit cycle bifurcation curve. (A) An overall perspective. (B) A detailed view of the cusp of the three-root solution space to describe the subregions.
	Figure 11. Exclusively Type 2 dynamics in the squid axon model.The time-course of the membrane voltage with increasing steady current for low and high K channel densities. (A)  = 1200 S/m2 and  = 50 S/m2. Onset frequency is 22 Hz. (B)  = 1200 S/m2 and  = 360 S/m2 (values used by Hodgkin and Huxley in their original study from 1952 [23]). The onset frequency is 52 Hz.
	Figure 12. Bifurcation curves and the three-root solution space for the myelinated axon model.Istim− diagrams at  = 1200 S/m2. The thick continuous line defines the region associated with three-root solutions of Equation 14. The thin continuous line is the Andronov-Hopf bifurcation curve and the hatched line is the double limit cycle bifurcation curve. (A) An overall perspective. (B) A detailed view of the cusp of the three-root solution space to describe the three subregions. The Bogdanov-Takens bifurcation point is marked.
	Figure 13. Oscillation maps for the axon models.Regions associated with oscillations in the − or − plane. (A) The frog myelinated axon model. (B) The squid giant axon model. As seen there is no C1a region in any of the maps and consequently both axon models lack Type 1 dynamics. Note also that the myelinated axon model (A) allows oscillations for  = 0 (no K channels). (C) Onset frequency in the myelinated axon model. (D) Onset frequency in the squid axon model. Circles indicates the original values used by Hodgkin and Huxley for the model of the axon of Loligo forbesi [23] and Frankenhaeuser and Huxley for model of the sciatic nerve of Xenpus leavis [22].

Amir, Burst discharge in primary sensory neurons: triggered by subthreshold oscillations, maintained by depolarizing afterpotentials. 2002, Pubmed

Amir, Burst discharge in primary sensory neurons: triggered by subthreshold oscillations, maintained by depolarizing afterpotentials. 2002, Pubmed
Arhem, Channel density regulation of firing patterns in a cortical neuron model. 2006, Pubmed
Arhem, Ion channel density and threshold dynamics of repetitive firing in a cortical neuron model. 2007, Pubmed
Bartos, Synaptic mechanisms of synchronized gamma oscillations in inhibitory interneuron networks. 2007, Pubmed
Bean, The action potential in mammalian central neurons. 2007, Pubmed
Brismar, Potential clamp analysis of membrane currents in rat myelinated nerve fibres. 1980, Pubmed
Cardin, Driving fast-spiking cells induces gamma rhythm and controls sensory responses. 2009, Pubmed
Desai, Plasticity in the intrinsic excitability of cortical pyramidal neurons. 1999, Pubmed
Desai, BDNF regulates the intrinsic excitability of cortical neurons. 1999, Pubmed
DODGE, Sodium currents in the myelinated nerve fibre of Xenopus laevis investigated with the voltage clamp technique. 1959, Pubmed , Xenbase
DODGE, Membrane currents in isolated frog nerve fibre under voltage clamp conditions. 1958, Pubmed
Ermentrout, The effects of spike frequency adaptation and negative feedback on the synchronization of neural oscillators. 2001, Pubmed
Ermentrout, Type I membranes, phase resetting curves, and synchrony. 1996, Pubmed
Fitzhugh, Impulses and Physiological States in Theoretical Models of Nerve Membrane. 1961, Pubmed
FRANKENHAEUSER, THE ACTION POTENTIAL IN THE MYELINATED NERVE FIBER OF XENOPUS LAEVIS AS COMPUTED ON THE BASIS OF VOLTAGE CLAMP DATA. 1964, Pubmed , Xenbase
Goldman, POTENTIAL, IMPEDANCE, AND RECTIFICATION IN MEMBRANES. 1943, Pubmed
Govaerts, The onset and extinction of neural spiking: a numerical bifurcation approach. 2005, Pubmed
Gutkin, Spike generating dynamics and the conditions for spike-time precision in cortical neurons. 2003, Pubmed
Hodgkin, The local electric changes associated with repetitive action in a non-medullated axon. 1948, Pubmed
HODGKIN, The effect of temperature on the electrical activity of the giant axon of the squid. 1949, Pubmed
HODGKIN, A quantitative description of membrane current and its application to conduction and excitation in nerve. 1952, Pubmed
Johansson, Membrane currents in small cultured rat hippocampal neurons: a voltage-clamp study. 1992, Pubmed
Kim, Regulation of dendritic excitability by activity-dependent trafficking of the A-type K+ channel subunit Kv4.2 in hippocampal neurons. 2007, Pubmed
Kole, Action potential generation requires a high sodium channel density in the axon initial segment. 2008, Pubmed
Kole, Is action potential threshold lowest in the axon? 2008, Pubmed
Kovalsky, Subthreshold oscillations facilitate neuropathic spike discharge by overcoming membrane accommodation. 2008, Pubmed
Lien, Kv3 potassium conductance is necessary and kinetically optimized for high-frequency action potential generation in hippocampal interneurons. 2003, Pubmed
Liu, Transition between two excitabilities in mesencephalic V neurons. 2008, Pubmed
Lundstrom, Two computational regimes of a single-compartment neuron separated by a planar boundary in conductance space. 2008, Pubmed
Marder, Modeling stability in neuron and network function: the role of activity in homeostasis. 2002, Pubmed
Marder, Variability, compensation and homeostasis in neuron and network function. 2006, Pubmed
MOORE, Resting and action potentials of the squid giant axon in vivo. 1960, Pubmed
Morris, Voltage oscillations in the barnacle giant muscle fiber. 1981, Pubmed
Prescott, Biophysical basis for three distinct dynamical mechanisms of action potential initiation. 2008, Pubmed
Prescott, Pyramidal neurons switch from integrators in vitro to resonators under in vivo-like conditions. 2008, Pubmed
Rojas-Piloni, Superficial dorsal horn neurons with double spike activity in the rat. 2007, Pubmed
Rush, The potassium A-current, low firing rates and rebound excitation in Hodgkin-Huxley models. 1995, Pubmed
Stemmler, How voltage-dependent conductances can adapt to maximize the information encoded by neuronal firing rate. 1999, Pubmed
St-Hilaire, Comparison of coding capabilities of Type I and Type II neurons. 2004, Pubmed
Stiefel, The effects of cholinergic neuromodulation on neuronal phase-response curves of modeled cortical neurons. 2009, Pubmed
Stiefel, Cholinergic neuromodulation changes phase response curve shape and type in cortical pyramidal neurons. 2008, Pubmed
Tateno, Rate coding and spike-time variability in cortical neurons with two types of threshold dynamics. 2006, Pubmed
Tateno, Phase resetting curves and oscillatory stability in interneurons of rat somatosensory cortex. 2007, Pubmed
Tateno, Threshold firing frequency-current relationships of neurons in rat somatosensory cortex: type 1 and type 2 dynamics. 2004, Pubmed
Traub, High-frequency population oscillations are predicted to occur in hippocampal pyramidal neuronal networks interconnected by axoaxonal gap junctions. 1999, Pubmed