Fiber Models and Activation

Fiber Models & Activation

Given the extracellular potential Vₑ sampled along each fiber and a stimulus pulse, golgi computes whether the fiber fires and at what amplitude — its activation threshold — using full biophysical cable models. Two interchangeable backends compute the same quantity at different speed/fidelity trade-offs.

Source: golgi/pipeline/fiber_sim.py (driver + dispatch), golgi/pipeline/fiber_backends.py (AxonML), golgi/state_defaults/fiber.py (defaults).

---

Two backends

Backend	Engine	Use
PyFibers / NEURON (default)	full Hodgkin–Huxley cable simulation in NEURON (Marshall et al. 2025; Hines & Carnevale 1997)	reference-grade thresholds; the validated path
AxonML (optional)	GPU surrogate of the MRG model (Torch)	high-throughput sweeps / populations; falls back to CPU without CUDA

Select the backend per run (backend="pyfibers" / "axonml"). AxonML is a separately-licensed, non-commercial Duke package that golgi never bundles — see License & Citation; the default NEURON/PyFibers path needs nothing extra.

Fiber models

• Myelinated fibers use the MRG model (McIntyre–Richardson–Grill double-cable; McIntyre et al. 2002). golgi offers both a continuous interpolated MRG (any diameter in range) and discrete MRG at the canonical McIntyre diameters.
• Unmyelinated (C) fibers use unmyelinated models (e.g. Tigerholm et al. 2014 / Sundt et al. 2015).

Diameter selects the model and its morphology (internodal spacing, node geometry). The diameter→model mapping and per-model ranges are configured in FIBER_MODEL_DIAMETER_CONFIG; the default fiber diameter is 5.7 µm with the interpolated MRG model. See Fiber Populations & Trajectories for how diameters are assigned across a population.

Pulses

Stimulus waveforms are built by build_pulse_waveform — monophasic or biphasic (cathodic- or anodic-first, with an optional interphase gap and charge balancing), parameterized by amplitude (mA), pulse width (µs), onset, and stop time. Pulse parameters are set in the single-fiber / population / sweep panels.

Thresholds & recruitment

An activation threshold is the smallest stimulus amplitude that produces a propagating action potential. golgi finds it by bisection (titration) on amplitude — the fiber is simulated at successive amplitudes, halving the bracket each time until the threshold is bracketed to a tolerance (default ±10 µA). Because the quasi-static field is linear in current, scaling amplitude simply scales Vₑ — no FEM re-solve is needed per amplitude (see Finite-Element Solver).

From thresholds across a population you get recruitment (fraction of fibers activated vs amplitude) and, per branch/type, fiber-type-selective recruitment — the inputs to selectivity analysis. golgi also reports conduction velocity (myelinated fibers conduct faster; saltatory vs continuous), which is one of the validation checks against Hursh and McIntyre references.

---

References

• McIntyre CC, Richardson AG, Grill WM (2002). Modeling the excitability of mammalian nerve fibers: influence of afterpotentials on the recovery cycle. J Neurophysiol 87:995. — the MRG myelinated-fiber model.
• Marshall DP, Farah ES, Musselman ED, Pelot NA, Grill WM (2025). PyFibers: an open-source NEURON-Python package to simulate responses of peripheral nerve fibers. PLOS Comput Biol. — the default fiber-simulation backend.
• Hines ML, Carnevale NT (1997). The NEURON simulation environment. Neural Comput 9:1179. — the simulator PyFibers runs on.
• Tigerholm J, et al. (2014). Modeling activity-dependent changes of axonal spike conduction in primary afferent neurons. J Neurophysiol. — unmyelinated C-fiber model.
• Sundt D, Gamper N, Jaffe DB (2015). Spike propagation through the dorsal root ganglia in an unmyelinated sensory neuron: a modeling study. J Neurophysiol. — unmyelinated C-fiber model.
• Hursh JB (1939). Conduction velocity and diameter of nerve fibers. Am J Physiol 127:131. — conduction-velocity validation target.

See Validation for the strength–duration and cross-platform threshold references.

---

See also

Fiber Populations & Trajectories · Recruitment Sweeps & Selectivity · Finite-Element Solver · Recording & CNAP · Validation