Neuron Signal Lab · Guide · glidepathscience.org
GuideNeuron Signal Lab
Contents
This Guide explains the ideas and equations behind every Walkthrough step, Concept Challenge question and Predict round. Part 1 prepares the Intro track: the membrane as a capacitor and a resistor, the axon as a leaky cable, the action potential, conduction speed and myelin. Part 2 prepares the Advanced track: the cable equation, the time course of a cable, Hodgkin and Huxley’s gates, how speed scales with diameter, and the myelinated cable. Each part takes about twelve minutes to read; the Reference collects common mistakes, the equations and further reading.
Part 1 · The essentials
Math you’ll need: algebra, Ohm’s law, the resistance of a wire, and one exponential, $e^{−x/λ}$ or $e^{−t/τ}$, as in a discharging capacitor; no calculus is required.
The membrane is an RC circuit
A neuron gathers signals on its dendrites and cell body and sends its own signal along a single long fiber, the axon, to terminals that pass it to the next cell. The signal starts near the cell body and travels out along the axon at a definite speed. To see how, start with the axon’s wall.
The cell membrane is a lipid bilayer about 5 nm thick: an electrical insulator with conducting salt water on both sides. An insulator between two conductors is a capacitor, and the membrane has a capacitance per unit area $C_m$ of about 1 µF/cm², which is 0.01 F/m². At rest the inside is about 65 mV negative relative to the outside, so a thin layer of negative charge lines the inside face and positive charge the outside. Moving the voltage by $ΔV$ takes a charge
$Q = C_mAΔV$
for a patch of area $A$. The membrane also leaks. Ion channels let a small current through all the time, so the membrane is a resistor too, with a specific resistance $R_m$ (ohms times area: a patch of area $A$ has resistance $R_m/A$). Resistor and capacitor sit side by side across the same patch, so a patch behaves like an RC circuit with the time constant
$τ_m = R_mC_m$
The area cancels: a bigger patch has more capacitance but proportionally more channels and less resistance, so every patch of one kind of membrane has the same $τ_m$. As in RC Circuit Lab, a steady current switched on into a patch charges it as $ΔV_∞(1 − e^{−t/τ_m})$, reaching 63 % of its final value at $t = τ_m$.
The leaky cable and its length constant
Inside the axon, current flows along the axoplasm, a salt solution whose resistivity $ρ_i$ is around 1 Ω·m, tens of millions of times copper’s. A length $L$ of an axon of diameter $d$ has the resistance of any wire,
$R = ρL/(πd^2/4)$
which for a thin axon is enormous: more than a gigaohm per millimeter at 1 µm. Put the pieces together and the axon becomes a ladder: the core’s resistance along the inside rail, and across each rung a patch of membrane, a resistor and a capacitor side by side, down to the outside, which is a good conductor. This is a cable, and the same model describes undersea telegraph cables, for which it was first worked out.
Inject a steady current at one end of a long passive cable and wait for it to settle. Each rung leaks a fraction of the current that reaches it, and the fraction is proportional to the voltage there, so the voltage falls by the same factor over every equal length: an exponential,
$V = V_0e^{−x/λ}$
The length constant λ is the distance over which the voltage falls to 1/e, 37 % of its starting value. It grows with a less leaky membrane and a wider core, and shrinks with a more resistive axoplasm:
$λ = \sqrt{R_md/(4ρ_i)}$
For typical axons λ is a fraction of a millimeter to a few millimeters. After five length constants less than 1 % of the voltage is left, and a nerve to the foot is about a meter long, hundreds of length constants: passive spread delivers nothing there. It is still how signals cross dendrites and cell bodies, where distances are short.
The action potential rebuilds itself
Real axons carry signals a meter without loss because their membrane does not just leak: it fires. The membrane holds voltage-gated sodium channels that open when the inside is depolarized past a threshold. Sodium ions rush in, driving the inside positive, up toward about +40 mV; within a millisecond the sodium channels close again (they inactivate) and potassium channels open, letting potassium out, so the voltage falls back, briefly below rest. That brief swing is the action potential, or spike.
On an axon, the current from a firing patch spreads ahead through the core exactly as in a passive cable, and it depolarizes the next patch past threshold, which fires in turn. The spike is rebuilt at every point, so it arrives at the far end at full size. It is all or none: a stimulus either starts a full spike or nothing, and a stronger stimulus makes neither a bigger nor a faster one. Behind the spike the membrane is refractory for a few milliseconds, which keeps the spike moving away from where it started.
Conduction speed
The speed of a nerve signal is measured as on any racetrack: two electrodes a known distance apart record when the spike arrives at each, and
$v = Δx/Δt$
Using the difference of the two times cancels the delay at the stimulus. A convenient fact: 1 mm/ms is 1 m/s. The squid’s giant axon, about half a millimeter across, conducts at about 20 m/s; thin unmyelinated fibers in our own bodies, a micrometer or so across, at under 2 m/s. What sets the speed is how far ahead the current from a firing patch reaches (λ) and how quickly it charges the membrane there: a wider axon spreads current farther and so conducts faster, and so does a less resistive axoplasm or a membrane with less capacitance. Warmth helps too, because the channels’ gates move faster.
Myelin, and what happens when it is lost
Vertebrates have a better way to go fast than growing giant axons. Glial cells wrap many axons in myelin, dozens or hundreds of turns of their own membrane, leaving short gaps about a millimeter apart: the nodes of Ranvier, where sodium channels are packed densely. With $n$ wraps, $2n + 1$ membranes are stacked in series: their resistances add, so the leak falls about $2n + 1$ times, and capacitors in series combine like resistors in parallel, so the capacitance falls the same factor. Between nodes, the axon becomes a well-insulated cable that takes very little charge to raise its voltage.
So the current from a firing node reaches the next node almost undiminished and charges it past threshold in a few microseconds. The spike is regenerated only at the nodes and seems to jump from one to the next: saltatory conduction, from the Latin saltare, to leap. In the sim a 10 µm axon conducts about 15 times faster with myelin than without.
Damage to myelin works the other way. In multiple sclerosis the body’s immune system attacks myelin in the brain and spinal cord. A thinned internode has more capacitance and leak to charge, so the next node fires later and the signal slows; a stretch with almost no myelin left soaks up the current a node needs, since it has the capacitance of bare membrane but very few sodium channels, and the signal can fail there altogether.
Part 2 · Going further
Math you’ll need: partial derivatives and a linear partial differential equation with its steady-state solutions, hyperbolic functions, the error function, and a system of nonlinear ordinary differential equations; dimensional reasoning for the scaling laws.
The cable equation and its steady states
Cable theory works per unit length. For an axon of diameter $d$, the membrane’s resistance times length is $r_m = R_m/(πd)$, the core’s resistance per length is $r_i = 4ρ_i/(πd^2)$, and the membrane’s capacitance per length is $c_m = C_mπd$. The axial current is $i = −(1/r_i)∂V/∂x$, where $V$ is the voltage across the membrane measured from rest. Kirchhoff’s current law at each point says that what the core loses leaves through the membrane, as capacitive current plus leak:
$(1/r_i)∂^2V/∂x^2 = c_m∂V/∂t + V/r_m$
Multiplying by $r_m$ gives the cable equation,
$λ^2∂^2V/∂x^2 − τ_m∂V/∂t − V = 0$, $λ = \sqrt{r_m/r_i}$, $τ_m = r_mc_m = R_mC_m$
In the steady state, $λ^2V″ = V$, with solutions $e^{±x/λ}$. A long cable fed at $x$ = 0 keeps only the decaying one, and its end presents the input resistance $R_∞ = r_iλ$, the core’s resistance over one length constant. A finite cable of length $L$ needs a condition at its far end. A sealed end lets no current out ($V′ = 0$), so $V ∝ \cosh((L − x)/λ)$ and
$R_{in} = R_∞\operatorname{coth}(L/λ)$
which is larger than $R_∞$: a short cable has less membrane to leak through. A killed end, held at rest, gives $V ∝ \sinh((L − x)/λ)$ and $R_{in} = R_∞\tanh(L/λ)$, smaller than $R_∞$. Neurons only a fraction of a length constant long are electrically compact: one electrode sees nearly the whole cell.
The time course of a cable
Switch a steady current on at the end of a long passive cable. A single patch would rise as $1 − e^{−t/τ_m}$, but the end of a cable shares its charge with its neighbors, which are charging too, and the solution of the cable equation at the injection site is
$V(0,t)/V_∞ = \mathrm{erf}(\sqrt{t/τ_m})$
At $t = τ_m$ this is erf(1) = 0.843, not 0.632: the injection site rises faster than a patch at first. Farther along, the rise is delayed and slower, so a cable both attenuates and smears out the timing of any input. Measuring curves like these, in space and in time, is how $λ$ and $τ_m$ are found on real axons; the sim’s Cable Bench shows both the 63 % and 84 % levels.
Hodgkin and Huxley’s gates
Hodgkin and Huxley described the squid membrane with three gating variables, each between 0 and 1. The sodium conductance is $g_{Na} = \bar g_{Na}m^3h$ and the potassium conductance $g_K = \bar g_Kn^4$, with $\bar g_{Na}$ = 120 and $\bar g_K$ = 36 mS/cm². Each gate relaxes toward a voltage-dependent value, $dx/dt = φ(α_x(1 − x) − β_xx)$, with rates fitted to their voltage-clamp data. The membrane equation is then
$C\,dV/dt = −\bar g_{Na}m^3h(V − E_{Na}) − \bar g_Kn^4(V − E_K) − g_L(V − E_L) + I$
The spike comes from the gates’ different speeds. $m$ is fast: once the voltage passes threshold, sodium activation runs away, and $V$ races toward $E_{Na}$ = +50 mV. $h$ (sodium inactivation) and $n$ (potassium activation) are slower; a little later they shut the sodium current off and turn the potassium current on, pulling $V$ down toward $E_K$ = −77 mV, a little past rest. Until $h$ recovers and $n$ falls back, the membrane is refractory: absolutely at first, then relatively, needing a larger stimulus. That wake is why a spike runs one way along an axon even though its current spreads both ways.
The rates depend on temperature through the factor $φ = 3^{(T − 6.3\ °\mathrm{C})/10}$ (a $Q_{10}$ of 3). A warmer membrane fires narrower spikes and conducts faster, up to a point: when the gates outrun the membrane’s charging, the spike shrinks and fails, which the sim shows for a bare squid-type axon above about 34 °C.
How speed scales with diameter
Since $r_m ∝ 1/d$ and $r_i ∝ 1/d^2$, the length constant grows as $λ ∝ \sqrt{d}$, while $τ_m$ does not depend on $d$ at all. A spike on a bare axon advances by about one length constant in a time set by the membrane, so
bare: $v ∝ \sqrt{d}$
and this is exact for the Hodgkin-Huxley cable: rescaling $x$ by $\sqrt{d}$ leaves its equations unchanged. In the sim at 18.5 °C, a bare fiber conducts at about 0.85 m/s at 1 µm, and the 476 µm squid axon at 18.6 m/s. Their equations solved precisely give 18.7 m/s (Hodgkin and Huxley’s hand calculation gave 18.8); the sim’s coarser grid runs about 1 % slow. To go ten times faster by size alone, a bare axon must be a hundred times wider; that is why squid grew giant axons.
A myelinated fiber scaled evenly, its internodes and its myelin growing with $d$, behaves differently. Each node-to-node trip takes about the same time whatever the size, while the distance between nodes grows with $d$, so
myelinated: $v ∝ d$
(Rushton, 1951). Large mammalian fibers follow this rule, at about 6 m/s per µm of outer diameter at body temperature (Hursh, 1939); the sim’s model fiber, with squid channels at 18.5 °C, gives about 4.2 m/s per µm of axon diameter.
The myelinated cable
Model a myelinated axon as a chain: passive internodes, cables with a tiny capacitance and leak, joined by short active nodes. A firing node drives current down the internode; because the internode is short compared with its own length constant and stores little charge, most of it reaches the next node within microseconds and charges that node’s small membrane past threshold. The ratio of the depolarization a node actually gets to the threshold it needs is the safety factor, typically several.
Internode length is a compromise. Longer internodes mean fewer stops, but each node must charge a longer, leakier stretch, so it reaches threshold later; in the sim, doubling the internodes of the 10 µm fiber from 1.4 mm to 2.8 mm makes it slightly slower, and much longer internodes fail. Near their real length, about a hundred fiber diameters, the speed changes only slowly with internode length, so a node can be moved or lost without a large change.
Demyelination attacks the safety factor directly. Thinning the myelin of one internode raises its capacitance and leak; the node after it charges more slowly and fires late, and the delay adds to every later arrival. With almost no myelin left, the bare stretch soaks up the current: the node after it never reaches threshold, and conduction is blocked. Comparing speeds shows why myelin matters so much: a bare axon would have to be $(v_{my}/v_1)^2$ micrometers wide to match a myelinated fiber.
Reference
Common mistakes
Part 1
- Treating the axon as a good wire: its core is tens of millions of times worse than copper, and passive spread fades within a few length constants.
- Thinking a nerve signal is a current flowing from one end to the other: each spike is ions crossing the membrane locally, rebuilt patch by patch.
- Mixing units: 1 µF/cm² is 0.01 F/m², and 1 µm² is 10⁻¹² m².
- Expecting τ to depend on the size of a patch: $τ_m = R_mC_m$ does not.
- Expecting a stronger stimulus to make a bigger or faster spike: it is all or none.
- Thinking myelin speeds the ions: it cuts the membrane’s leak and capacitance between nodes, so the spike jumps from node to node.
Part 2
- Using a patch’s 63 % for the end of a cable: the injection site follows erf(√(t/τ)), 84 % at $τ_m$.
- Confusing the sealed end’s $\operatorname{coth}$ with the killed end’s $\tanh$.
- Using $v ∝ d$ for bare axons or $v ∝ \sqrt{d}$ for myelinated ones.
The equation sheet
| Part 1 | |
|---|---|
| the membrane | $Q = C_mAΔV$; $C_m$ ≈ 1 µF/cm² = 0.01 F/m² |
| its time constant | $τ_m = R_mC_m$ |
| the core | $R = ρL/(πd^2/4)$ |
| the cable | $λ = \sqrt{R_md/(4ρ_i)}$; $V = V_0e^{−x/λ}$ |
| speed | $v = Δx/Δt$; $t = x/v$ |
| Part 2 | |
|---|---|
| per length | $r_m = R_m/(πd)$, $r_i = 4ρ_i/(πd^2)$, $λ = \sqrt{r_m/r_i}$, $R_∞ = r_iλ$ |
| the cable equation | $λ^2∂^2V/∂x^2 − τ_m∂V/∂t − V = 0$ |
| finite cables | sealed: $R_{in} = R_∞\operatorname{coth}(L/λ)$; killed: $R_∞\tanh(L/λ)$ |
| the time course | $V(0,t)/V_∞ = \mathrm{erf}(\sqrt{t/τ_m})$ |
| scaling | bare: $v ∝ \sqrt{d}$; myelinated: $v ∝ d$ |
| temperature | $φ = 3^{(T − 6.3\ °\mathrm{C})/10}$ |
Symbols: $C_m$ and $R_m$ the membrane’s capacitance and resistance per area (times area, for $R_m$); $ρ_i$ the axoplasm’s resistivity; $d$ the axon’s diameter; $λ$ the length constant and $τ_m$ the membrane time constant; $r_m$, $r_i$, $c_m$ the same per unit length; $R_∞$ a long cable’s input resistance; $L$ a cable’s length; $v$ the conduction speed; $m$, $h$, $n$ Hodgkin and Huxley’s gates; $φ$ the temperature factor.
Sources and further reading
- OpenStax, Biology 2e, Chapter 35, “The Nervous System,” section 35.2 “How Neurons Communicate,” openstax.org (CC BY-NC-SA 4.0): the resting potential, the action potential and myelin in a biology course’s words.
- A. L. Hodgkin and A. F. Huxley, J. Physiol. 117, 500–544 (1952), the model this sim uses; W. A. H. Rushton, J. Physiol. 115, 101–122 (1951), on the effects of fiber size in medullated (myelinated) nerve; J. B. Hursh, Am. J. Physiol. 127, 131–139 (1939), on speed and diameter.
- This sim’s Field Manual (H) for the model, its typical values and its limits; RC Circuit Lab for one RC element; Diffusion & Osmosis Lab for the ion gradients behind the resting potential.
Neuron Signal Lab v1.0 · © 2026 Dr. David Meier · CC BY-NC-SA 4.0 · glidepathscience.org