17.09.02 · mol-cell-bio / cellular-neuroscience

The action potential — ionic basis

draft3 tiersLean: none

Anchor (Master): Hille, *Ion Channels of Excitable Membranes* (3rd ed., Sinauer 2001) — the canonical reference; Koch, *Biophysics of Computation* (Oxford University Press 1999), Chs. 6-7; Kandel et al., *Principles of Neural Science* (5th ed.) — advanced sections including patch-clamp and Markov-state modelling; Hodgkin & Huxley 1952 (*J. Physiol.* 117, 500-544) — the four foundational papers culminating in the quantitative description; Neher & Sakmann 1976 (*Nature* 260, 799-802) — patch-clamp recording of single-channel currents; Sakmann & Neher 1995 *Single-Channel Recording* (2nd ed., Plenum); Izhikevich, *Dynamical Systems in Neuroscience* (MIT Press 2007) — bifurcation analysis of HH and reduced models

Intuition [Beginner]

Neurons have one fundamental job: get a signal from one end of a long, thin cell to the other quickly and reliably, then pass it on. The signal in question — the action potential — is a sharp electrical pulse that propagates down the axon at speeds between roughly one and one hundred metres per second, depending on the neuron's size and whether the axon is wrapped in myelin insulation. A typical pulse lasts about a millisecond and changes the voltage across the cell membrane by about a hundred millivolts before snapping back. The whole structure of the nervous system is built on millions of these pulses per second running across neurons that are linked together at synapses.

The pulse is not built by sending electrons down a wire the way a copper cable does. Cells are full of salt water on the inside and surrounded by salt water on the outside; the wire would short instantly. Instead, the pulse is built by charged ions moving across the cell membrane through specialised protein channels. The four ions that matter most are sodium ( $Na^{+}$ ), potassium ( $K^{+}$ ), chloride ( $Cl^{-}$ ), and calcium ( $Ca^{2 +}$ ). The cell membrane is a lipid bilayer that ions cannot cross on their own; the protein channels are the only paths through, and the cell controls those paths.

At rest, the cell uses ATP-driven pumps to push sodium out and potassium in, building up unequal concentrations: lots of sodium outside and very little inside, lots of potassium inside and a little outside. Because each ion carries a charge, moving them around moves charge around, and at equilibrium the inside of the cell sits at a resting potential of about $- 70$ millivolts relative to the outside — a steady-state imbalance maintained by continuous pumping against passive leak.

The unequal concentrations store potential energy in the same way a stretched spring does. Energy gets stored as separation — chemical separation here, not mechanical. When sodium-permeable channels open, sodium ions rush in along their concentration gradient and along the electrical gradient that pulls positive charges toward the negative inside. Both gradients point the same way; the rush is fast.

The action potential is the choreographed opening and closing of those channels. The sequence is rigid. A small initial depolarisation (the inside becoming less negative) triggers sodium channels to open, sodium pours in, and the inside swings positive in a fraction of a millisecond — that is the upstroke of the spike.

Then the sodium channels close themselves automatically through a separate mechanism (inactivation) at almost the same moment that potassium channels open. Potassium flows out, taking positive charge with it, and the inside swings back negative — the repolarisation. The potassium channels are slower to close than they were to open, so the inside actually overshoots its resting value briefly and sits a few millivolts below $- 70$ for a moment — the undershoot or afterhyperpolarisation. Then everything resets to the resting state, ready for the next spike.

Two features of this signal matter most.

First, it is all-or-none. A depolarisation that does not reach a critical threshold (about $- 55$ mV in many neurons) produces no spike at all; the channels never start their cascade. A depolarisation that crosses the threshold produces a full spike, identical in size and shape every time, regardless of how far the initial push exceeded the threshold. The cell does not signal how strong the stimulus was by making bigger spikes; it signals strength by making spikes more often. This is sometimes called a digital code at the spike level, in contrast to the analog graded potentials in dendrites and at synapses where signal size does scale with input.

Second, it propagates. As one patch of membrane spikes, it pulls current from the neighbouring patch, depolarising it past threshold, which spikes in turn. The pulse moves down the axon without losing amplitude — the membrane regenerates the signal at every step. In myelinated axons the pulse leaps from one bare patch of membrane (the node of Ranvier) to the next, skipping the insulated stretches; this is called saltatory conduction and is why myelinated axons conduct fifty times faster than equivalent unmyelinated ones.

So a neuron is, viewed from this scale, a cell that converts incoming graded signals into a digital train of identical pulses and ships them down a long thin process to the next neuron. The action potential is the unit of that conversion. Every act of nervous-system communication — every twitch, every percept, every decision — runs through stacks of these pulses generated by gating proteins that flip open and shut in milliseconds.

Visual [Beginner]

The canonical picture is a voltage trace recorded from inside an axon over a few milliseconds. The trace sits flat at the resting potential of about $- 70$ mV, jumps almost vertically up to a peak near $+ 40$ mV in about half a millisecond, falls back through $- 70$ mV in another millisecond, undershoots by ten or fifteen millivolts, and slowly returns to rest. Below the voltage trace, a second pair of curves shows the sodium conductance and potassium conductance: the sodium conductance rises first and falls quickly; the potassium conductance rises later and falls more slowly. The two together reproduce the spike.

$Idealised action potential. Top: membrane voltage $V_m(t)$ rising from a resting $-70$ mV through threshold near $-55$ mV, peaking around $+40$ mV at the spike apex, falling back through the resting potential and undershooting briefly. Bottom: ionic conductances $g_{\rm Na}(t)$ (rises early and falls fast — sodium influx drives the upstroke) and $g_K(t)$ (rises later and falls more slowly — potassium efflux drives repolarisation and undershoot). The two conductances together generate the spike.$

A second useful picture is the membrane as a tiny battery in parallel with a tiny capacitor. The battery is the ionic equilibrium; the capacitor is the lipid bilayer (about one microfarad per square centimetre, a quantity Hodgkin and Huxley measured first). Charging and discharging the capacitor as ions flow through opened channels is exactly what the voltage trace records.

Worked example [Beginner]

Start with the resting cell: inside negative, outside positive, sodium concentrated outside, potassium concentrated inside, everything stable.

Step 1. A small depolarising current — say, from an excitatory synapse upstream — nudges the membrane voltage from $- 70$ mV up to $- 60$ mV, then $- 55$ mV.

Step 2. At $- 55$ mV (the threshold), voltage-gated sodium channels start opening. Each channel, when open, lets sodium ions through at a rate determined by the difference between the current voltage and the sodium equilibrium potential ( $+ 60$ mV). The driving force is huge: sodium wants in.

Step 3. Sodium rushes in. Each ion carries one positive charge across the membrane. The capacitor charges up — toward the positive direction. The voltage rises. As it rises, more sodium channels open (this is the positive feedback that makes the spike steep). The voltage shoots from $- 55$ mV to $+ 30$ mV in about half a millisecond.

Step 4. Two things stop the upstroke. First, the sodium channels inactivate themselves: a second gate inside the channel swings shut, blocking flow even though the activation gate is still open. Second, voltage-gated potassium channels — which respond more slowly to the depolarisation — open. Potassium starts flowing out.

Step 5. Potassium pours out, carrying positive charge to the outside, repolarising the cell. The voltage drops back through zero, then back to $- 70$ , then briefly past it. Why the overshoot? The potassium channels close slowly, so they keep letting potassium out for a few milliseconds after the voltage hits the resting value. The cell sits at about $- 85$ mV (the potassium equilibrium potential) until the potassium channels finally close.

Step 6. Sodium pumps and potassium-channel closure restore the resting state. The cell is now ready for another spike, except that for a millisecond or two the sodium channels are still inactivated and cannot reopen — this refractory period is what enforces a maximum spiking rate and forces the action potential to propagate forward rather than backward along the axon.

What this tells us: the action potential is a stereotyped sequence of channel openings and closings, choreographed by the voltage itself, that uses stored chemical energy (the unequal ion distributions maintained by pumps) to push a millivolt-scale electrical pulse cleanly along an axon. Nothing about the spike is mysterious once you accept that the membrane has channels and the channels know how to respond to voltage.

Check your understanding [Beginner]

Exercise (easy, multiple choice).

At rest, inside a typical neuron there is roughly $10$ times more potassium than outside, and outside there is roughly $10$ times more sodium than inside. If a channel opens that lets only sodium ions through, which direction will sodium flow at the moment the channel opens, while the membrane is still near $- 70$ mV?

A. From inside to outside, driven by the concentration gradient. B. From outside to inside, driven by both the concentration gradient and the electrical pull on a positive charge into a negative interior. C. Sodium will not move because the membrane is at equilibrium. D. From inside to outside, driven by the electrical gradient.

Hint

Sodium is more concentrated outside and the inside of the cell is negative. Which way do these two forces push a positively-charged sodium ion?

Answer

B. Both forces push sodium inward: the concentration gradient wants to equalise the higher outside concentration with the lower inside one (down the gradient is inward), and the inside is negatively charged so it electrically attracts a positive ion. The two driving forces aligned is what makes the sodium influx during the spike so fast and steep.

Exercise (easy, multiple choice).

A myelinated axon conducts an action potential roughly $50$ times faster than an unmyelinated axon of the same diameter. The mechanism is:

A. Myelin enables electrons to flow directly through the axoplasm. B. Myelin insulates stretches of axon and forces the spike to regenerate only at gaps called nodes of Ranvier, so the pulse leaps from node to node. C. Myelin replaces sodium with a faster ion species. D. Myelin allows the membrane to operate at a higher temperature.

Hint

The technical term is saltatory conduction, from the Latin saltare (to leap).

Answer

B. Myelin wraps the axon with many layers of insulating membrane, leaving only periodic gaps (nodes of Ranvier) where voltage-gated channels are concentrated. The depolarisation produced at one node spreads passively along the insulated stretch — fast, because the insulation reduces current leak — and triggers regeneration at the next node. The spike "leaps" from node to node rather than regenerating continuously, which is faster.

Exercise (medium, numeric).

A neuron fires at a steady rate of $40$ spikes per second in response to a strong sustained input, where each spike lasts about $1.5$ milliseconds. Roughly what fraction of the time is the cell actually in the active-spike phase, as opposed to the rest of the inter-spike interval?

Hint

Compute spike-duration times spike-rate.

Answer

Roughly $0.06$ , or $6%$ . At $40$ Hz, the inter-spike interval is $25$ ms; the spike itself occupies $1.5$ ms of each interval. So the cell is in the spike phase only $1.5/25 = 6%$ of the time. The rest is the rest, the post-spike refractory recovery, and the slow climb back up to threshold under continued input. Coding theories that use spike-timing as the message do not care about how much time the spikes themselves occupy; they care about the precise moments at which the spike phase starts.

Formal definition [Intermediate+]

The membrane of a neuron separates two aqueous compartments with different ionic compositions. For the squid giant axon studied by Hodgkin and Huxley, typical intracellular concentrations are $[K^{+}]_{i} \approx 400$ mM, $[Na^{+}]_{i} \approx 50$ mM, $[Cl^{-}]_{i} \approx 40$ mM, $[Ca^{2 +}]_{i} \approx 1 0^{- 4}$ mM; extracellular concentrations are $[K^{+}]_{o} \approx 20$ mM, $[Na^{+}]_{o} \approx 440$ mM, $[Cl^{-}]_{o} \approx 560$ mM, $[Ca^{2 +}]_{o} \approx 10$ mM. These ratios are maintained by the Na+/K+ ATPase pump at a metabolic cost of roughly one ATP per three sodium ions extruded.

The membrane potential $V_{m} := V_{in} - V_{out}$ is the voltage across the bilayer, conventionally with respect to the outside as ground. At rest in a mammalian neuron, $V_{m} \approx - 70$ mV.

The Nernst equilibrium potential for an ion species $X$ with charge $z_{X}$ and concentrations $[X]_{o}$ , $[X]_{i}$ is the value of $V_{m}$ at which the electrical force on $X$ exactly cancels its chemical-gradient force, so the net flux of $X$ through a permeable membrane is zero. From the canonical-ensemble Boltzmann factor 11.04.01 pending applied to an ion in either compartment at temperature $T$ — the probability ratio of finding the ion at electrochemical potential $μ_{o} = k_{B} T ln [X]_{o} + z_{X} e V_{out}$ vs $μ_{i} = k_{B} T ln [X]_{i} + z_{X} e V_{in}$ equals the Boltzmann factor $exp (- (μ_{i} - μ_{o}) / k_{B} T)$ — equilibrium requires $μ_{i} = μ_{o}$ , giving

V_{eq} (X) = \frac{R T}{z _{X} F} ln \frac{[ X ] _{o}}{[ X ] _{i}}

where $R$ is the gas constant, $F$ is Faraday's constant, and we have rewritten in molar units ( $k_{B} N_{A} = R$ , $e N_{A} = F$ ). At $T = 310$ K (mammalian body temperature), $R T / F \approx 26.7$ mV and $(R T / F) ln 10 \approx 61.5$ mV. The typical equilibrium potentials computed from squid concentrations and body temperature are $V_{eq} (Na^{+}) \approx + 55$ mV, $V_{eq} (K^{+}) \approx - 75$ mV, $V_{eq} (Cl^{-}) \approx - 70$ mV, $V_{eq} (Ca^{2 +}) \approx + 120$ mV.

When the membrane is permeable to several ions simultaneously, the resting potential is set not by any single Nernst potential but by a permeability-weighted combination. The Goldman-Hodgkin-Katz equation for the resting potential under the constant-field assumption is

V_{m} = \frac{R T}{F} ln \frac{P _{K} [ K ^{+} ] _{o} + P _{Na} [ Na ^{+} ] _{o} + P _{Cl} [ Cl ^{-} ] _{i}}{P _{K} [ K ^{+} ] _{i} + P _{Na} [ Na ^{+} ] _{i} + P _{Cl} [ Cl ^{-} ] _{o}},

where $P_{X}$ is the permeability of the resting membrane to ion species $X$ . At rest, $P_{K} ≫ P_{Na} ≫ P_{Cl}$ , so $V_{m}$ sits close to the potassium Nernst potential. During the upstroke of an action potential the sodium permeability rises by orders of magnitude and $V_{m}$ swings toward the sodium Nernst potential. The peak of the spike approaches $V_{eq} (Na^{+})$ but typically does not reach it — repolarising potassium current and sodium inactivation kick in first.

The membrane is treated electrically as a parallel combination of a capacitor (the lipid bilayer, with specific capacitance $C_{m} \approx 1$ μF/cm²) and ionic conductances in series with their respective batteries. Kirchhoff's current law at the inner membrane face gives

C_{m} \frac{d V _{m}}{d t} = - X \sum g_{X} (V_{m}, t) (V_{m} - V_{eq} (X)) + I_{ext} (t)

where $g_{X}$ is the conductance per unit area of ion $X$ and $I_{ext}$ is any externally injected current. The signs are set so that an ionic current is positive when it flows from inside to outside (i.e., the conventional current direction for cations leaving the cell), which makes the right-hand side a driving force multiplied by a conductance in the natural way.

The Hodgkin-Huxley model specialises this to the squid giant axon with three currents — voltage-gated sodium, voltage-gated potassium (the delayed rectifier), and an unspecified leak ^{[Hodgkin & Huxley 1952]}:

C_{m} \frac{d V _{m}}{d t} = - \overset{g}{ˉ}_{Na} m^{3} h (V_{m} - E_{Na}) - \overset{g}{ˉ}_{K} n^{4} (V_{m} - E_{K}) - \overset{g}{ˉ}_{L} (V_{m} - E_{L}) + I_{ext},

with maximum conductances $\overset{g}{ˉ}_{Na}, \overset{g}{ˉ}_{K}, \overset{g}{ˉ}_{L}$ and reversal potentials $E_{Na} = V_{eq} (Na^{+})$ , $E_{K} = V_{eq} (K^{+})$ , $E_{L}$ chosen to fix the resting potential. The dimensionless gating variables $m, h, n \in [0, 1]$ obey first-order kinetics

\frac{d m}{d t} = α_{m} (V_{m}) (1 - m) - β_{m} (V_{m}) m, \frac{d h}{d t} = α_{h} (V_{m}) (1 - h) - β_{h} (V_{m}) h, \frac{d n}{d t} = α_{n} (V_{m}) (1 - n) - β_{n} (V_{m}) n,

with empirical voltage-dependent rate functions $α_{x}, β_{x}$ that Hodgkin and Huxley fit from voltage-clamp data on the squid axon. The interpretation: $m$ is the sodium activation (probability that an activation gate is open; the channel needs three of them, hence $m^{3}$ ), $h$ is the sodium inactivation (probability the inactivation gate is open), and $n$ is the potassium activation (the potassium channel needs four open gates, hence $n^{4}$ ). The full state of the system at any time is the four-tuple $(V_{m}, m, h, n) \in R \times [0, 1]^{3}$ , and Hodgkin-Huxley's equations are a four-dimensional flow on this domain.

Counterexamples to common slips

The resting potential is not the Nernst potential of any single ion in general. It is the GHK-weighted average of the relevant ions' equilibrium potentials, weighted by membrane permeabilities. It sits closest to $V_{eq} (K^{+})$ because $P_{K}$ dominates at rest, but it is not equal to it; the difference is exactly what the Na+/K+ ATPase has to compensate for.
The all-or-none principle is a property of the spike once threshold is crossed, not of the entire response. Subthreshold depolarisations are graded and decay; only the regenerative part above threshold is stereotyped.
$g_{X}$ is the conductance per unit area, not the conductance of a single channel. A single channel is a stochastic gate flipping open and shut; $g_{X} (V_{m}, t)$ is the population-averaged conductance, justified at the macroscopic scale by the law of large numbers across thousands or millions of channels per square micron.
The action potential does not "carry" charge from one end of the axon to the other. The ions that enter at one location stay near that location; the action potential is a regenerative wave in which the local ionic transients reignite the same transients in the neighbouring patch. The signal propagates; the matter does not.
The Hodgkin-Huxley model is non-relativistic, non-quantum, classical. The metalloprotein chemistry inside individual ion channels is genuinely quantum-mechanical (selectivity-filter coordination chemistry, partial-charge interactions); the macroscopic channel-population description averages this out and treats the channel as a stochastic gate with phenomenological rate constants.

Key theorem with proof [Intermediate+]

Theorem (Nernst equation from the canonical ensemble). Let an ion species $X$ with charge $z_{X} e$ be distributed between two compartments at temperature $T$ , separated by a membrane permeable to $X$ but to no other species. Let the inside compartment be at electrical potential $V_{in}$ and the outside at $V_{out}$ , with $V_{m} := V_{in} - V_{out}$ . At thermodynamic equilibrium, the ratio of concentrations satisfies $[X]_{o} / [X]_{i} = exp (z_{X} F V_{m} / R T)$ , equivalently

V_{m} = \frac{R T}{z _{X} F} ln \frac{[ X ] _{o}}{[ X ] _{i}} .

Proof. At fixed $T$ , the canonical ensemble 11.04.01 pending gives the probability density for an ion in a small volume around position $r$ as proportional to $exp (- β ε (r))$ where $β = 1/ k_{B} T$ and $ε (r)$ is the single-ion energy at $r$ . Inside each compartment the energy is dominated by the local electrostatic potential, so the ion density on the inside is $ρ_{i} \propto exp (- β z_{X} e V_{in})$ and on the outside $ρ_{o} \propto exp (- β z_{X} e V_{out})$ , with the same proportionality constant (the ion is the same species, the kinetic-energy integral over momenta cancels). The concentration ratio at equilibrium is therefore

\frac{[ X ] _{o}}{[ X ] _{i}} = \frac{ρ _{o}}{ρ _{i}} = exp (- β z_{X} e (V_{out} - V_{in})) = exp (β z_{X} e V_{m}) .

Taking logarithms and rewriting in molar units via $k_{B} N_{A} = R$ and $e N_{A} = F$ :

ln \frac{[ X ] _{o}}{[ X ] _{i}} = \frac{z _{X} F V _{m}}{R T}, i.e., V_{m} = \frac{R T}{z _{X} F} ln \frac{[ X ] _{o}}{[ X ] _{i}} . ■

The proof exploits exactly two pieces of structure: the canonical-ensemble Boltzmann factor (the entire stat-mech foundation of 11.04.01 pending) and the fact that the membrane lets one species cross so equilibrium for that species can be reached. When the membrane is permeable to multiple species, no single-species equilibrium can be achieved (Donnan equilibrium notwithstanding); the steady-state membrane potential is then set by the GHK equation rather than by any one Nernst potential.

Corollary (driving force). The current per unit area carried by ion $X$ across a membrane patch with conductance $g_{X}$ is $I_{X} = g_{X} (V_{m} - V_{eq} (X))$ , with $V_{m} - V_{eq} (X)$ called the driving force.

The driving force vanishes when $V_{m} = V_{eq} (X)$ , which is the operational meaning of the Nernst equation: the equilibrium potential is the value of $V_{m}$ at which the ion- $X$ current changes sign and the channel- $X$ flow rate is zero on average. Above $V_{eq} (X)$ , the channel carries current of one sign; below, of the other. This makes voltage-clamp identification of ion species across a current trace unambiguous: identify the species $X$ such that the current reverses when the clamped voltage is set to $V_{eq} (X)$ .

Worked example at intermediate level: the resting potential from GHK

Take a model neuron with $P_{K} : P_{Na} : P_{Cl} = 1 : 0.04 : 0.45$ and the squid concentrations $[K^{+}]_{i} = 400$ , $[K^{+}]_{o} = 20$ , $[Na^{+}]_{i} = 50$ , $[Na^{+}]_{o} = 440$ , $[Cl^{-}]_{i} = 40$ , $[Cl^{-}]_{o} = 560$ (all in mM). At $T = 293$ K, $R T / F = 25.3$ mV. The GHK equation gives

V_{m} = 25.3 mV \cdot ln \frac{1 \cdot 20 + 0.04 \cdot 440 + 0.45 \cdot 40}{1 \cdot 400 + 0.04 \cdot 50 + 0.45 \cdot 560} = 25.3 \cdot ln \frac{55.6}{654} \approx - 62 mV,

close to the measured resting potential of the squid giant axon ( $- 65$ mV; the residual discrepancy is the active contribution of the Na+/K+ pump, which is electrogenic). Notice that $V_{m}$ sits much closer to $V_{eq} (K) = - 75$ mV than to $V_{eq} (Na) = + 54$ mV — exactly because $P_{K}$ is twenty-five times larger than $P_{Na}$ at rest. During the spike, $P_{Na}$ transiently rises by three orders of magnitude and the GHK weighted average flips its allegiance to sodium.

Exercises [Intermediate+]

Exercise 2 (easy, multiple choice).

A voltage-clamped patch is held at $V_{m} = - 50$ mV. The sodium reversal potential is $V_{eq} (Na^{+}) = + 60$ mV. If voltage-gated sodium channels are open, the sodium current $I_{Na}$ at this clamped voltage flows:

A. Outward, carrying positive charge out. B. Inward, carrying positive charge in. C. Zero, because the channel is closed. D. Outward, carrying positive charge in.

Hint

Compute the driving force $V_{m} - V_{eq} (Na^{+})$ .

Answer

B. Driving force $V_{m} - V_{eq} (Na^{+}) = - 50 - 60 = - 110$ mV; conductance $g_{Na} > 0$ . So $I_{Na} = g_{Na} \cdot (- 110 mV) < 0$ . By the sign convention (positive current = positive charge flowing outward, from inside to outside), $I_{Na} < 0$ means sodium flows inward, carrying positive charge into the cell. This is the upstroke direction of the action potential.

Exercise 4 (medium, symbolic).

For a one-ion membrane with conductance $g$ and reversal potential $E$ , write down the linear single-ODE Hodgkin-Huxley reduction (no gating variables: constant $g$ , constant $E$ , plus capacitance $C_{m}$ and constant injected current $I$ ). Solve for $V_{m} (t)$ given initial condition $V_{m} (0) = V_{0}$ .

Hint

The equation $C_{m} d V / d t = - g (V - E) + I$ is linear and first-order. The steady-state is $V_{\infty} = E + I / g$ .

Answer

$C_{m} \dot{V} = - g (V - E) + I$ rewrites as $\dot{V} = - (V - V_{\infty}) / τ$ with $τ := C_{m} / g$ and $V_{\infty} := E + I / g$ . The solution is

V_{m} (t) = V_{\infty} + (V_{0} - V_{\infty}) e^{- t / τ} .

The membrane is a low-pass filter on injected current with time constant $τ$ , the membrane time constant. For a passive neuron with $C_{m} = 1$ μF/cm² and $g_{L} = 0.3$ mS/cm², $τ \approx 3$ ms. This determines how rapidly the cell can integrate synaptic inputs — inputs separated by more than $τ$ are effectively independent; those within $τ$ summate.

Exercise 5 (medium, symbolic).

Identify the dynamical role of each Hodgkin-Huxley gating variable. Specifically: which variable controls the rising edge of the spike, which controls the falling edge, and which is responsible for the refractory period?

Hint

$m$ activates fast; $h$ inactivates slow; $n$ activates slow.

Answer

$m$ — sodium activation — rises rapidly above threshold and is responsible for the rising edge (the positive-feedback explosion that pulls $V_{m}$ toward $E_{Na}$ ). $n$ — potassium activation — rises more slowly and drives the falling edge (potassium efflux repolarising the cell as $g_{K}$ rises). $h$ — sodium inactivation — falls during the upstroke and is responsible for the refractory period: even after $V_{m}$ has returned to rest, $h$ recovers slowly, so a fraction of the sodium channels remain inactivated for a few milliseconds and threshold is harder to reach again. The interaction of fast $m$ activation, slow $h$ inactivation, and slow $n$ activation is what makes the spike both sharp and self-limiting.

Exercise 6 (medium, numeric).

A patch of squid axon with $\overset{g}{ˉ}_{Na} = 120$ mS/cm², $E_{Na} = + 50$ mV, is held by a voltage clamp at $V_{m} = 0$ mV with the sodium channels maximally activated (so $m^{3} h \approx 1$ ). What is the sodium current density at this clamped voltage?

Hint

$I_{Na} = \overset{g}{ˉ}_{Na} m^{3} h (V_{m} - E_{Na})$ .

Answer

$I_{Na} = 120 mS/cm^{2} \cdot 1 \cdot (0 - 50) mV = - 6000 μ A/cm^{2} = - 6 mA/cm^{2}$ . The negative sign indicates inward current (positive charge moving into the cell) — the upstroke direction. At maximum activation during a spike, the sodium current is in the milliamp-per-square-centimetre range, which charges the $1 μ F/cm^{2}$ membrane capacitance at a rate $d V / d t = I / C_{m} = 6000$ V/s. Note that the idealisation $m^{3} h \approx 1$ overstates the actual gating: during the upstroke, $m^{3} h \sim 0.1$ – $0.3$ because $h$ -inactivation begins before $m$ -activation peaks, giving a realistic slope of roughly $600$ – $1800$ V/s ( $0.6$ – $1.8$ mV/ $μ$ s), in agreement with the observed $200$ – $500$ mV/ms upstroke slopes.

Exercise 7 (hard, symbolic).

Take a simplified two-variable reduction of HH: $V_{m}$ and a single recovery variable $w$ (treating $m$ as instantaneously slaved to $V_{m}$ and $h$ as approximately constant on the spike timescale). Show that the resulting system has the FitzHugh-Nagumo form

\dot{V} = V - V^{3} /3 - w + I, \overset{w}{˙} = ϵ (V + a - b w),

after appropriate rescaling. Identify each Hodgkin-Huxley element with its FitzHugh-Nagumo counterpart.

Hint

The cubic comes from the $V$ -dependence of the steady-state $m_{\infty}^{3}$ near threshold, expanded to leading nontrivial order. $w$ represents the combination $n + (1 - h)$ , the slow refractory recovery.

Answer

Adiabatic elimination of $m$ replaces $m (t)$ by $m_{\infty} (V_{m})$ , the steady-state activation curve. Near threshold, $m_{\infty} (V)^{3} (V - E_{Na})$ has the qualitative shape of a cubic in $V$ with a region of negative slope (a region of negative differential conductance), which is the mechanism of regenerative excitability. Combining the sodium inactivation $h$ with potassium activation $n$ into a single slow recovery variable $w$ (since both inhibit the spike), rescaling $V_{m}$ around the resting potential, and rescaling time by the membrane time constant produces the FitzHugh-Nagumo system as written, with $ϵ ≪ 1$ encoding the time-scale separation between $V$ -dynamics (fast) and $w$ -dynamics (slow). The fast nullcline $\dot{V} = 0$ is the cubic $w = V - V^{3} /3 + I$ ; the slow nullcline $\overset{w}{˙} = 0$ is the line $V = b w - a$ . The intersection of these nullclines is the equilibrium; whether the system spikes depends on whether the intersection lies on the middle (negative-slope) branch of the cubic (unstable → limit cycle, spiking) or on the outer branches (stable). The threshold-to-spike transition is exactly the Hopf bifurcation through which the equilibrium loses stability as $I$ varies.

Exercise 8 (hard, symbolic).

In FitzHugh-Nagumo with $\dot{V} = V - V^{3} /3 - w + I$ and $\overset{w}{˙} = ϵ (V + a - b w)$ , identify the bifurcation that occurs as $I$ is varied from below threshold (cell silent) to above threshold (cell repetitively spiking). Compute the critical $I$ value and the type of bifurcation.

Hint

Find the fixed point $(V^{*}, w^{*})$ as a function of $I$ , compute the linearisation, and look for purely imaginary eigenvalues.

Answer

The fixed point satisfies $w^{*} = V^{*} - (V^{*})^{3} /3 + I$ and $w^{*} = (V^{*} + a) / b$ ; eliminate $w^{*}$ to get an implicit cubic for $V^{*}$ in $I$ . The Jacobian at the fixed point is

J = (1 - (V^{*})^{2} ϵ - 1 - ϵ b) .

Trace $= 1 - (V^{*})^{2} - ϵ b$ and determinant $= - ϵ b (1 - (V^{*})^{2}) + ϵ = ϵ (b (V^{*})^{2} - b + 1)$ . A Hopf bifurcation occurs when the trace vanishes (purely imaginary eigenvalues), requiring $(V^{*})^{2} = 1 - ϵ b$ , with positive determinant. Substituting back into the fixed-point equation gives the critical $I_{Hopf}$ as a function of $a, b, ϵ$ . For the canonical Bonhoeffer-Van der Pol parameters $a = 0.7$ , $b = 0.8$ , $ϵ = 0.08$ , Hopf bifurcation occurs near $I \approx 0.33$ , in good agreement with the location at which the system transitions from silent (stable spiral) to repetitively spiking (stable limit cycle). The bifurcation is supercritical for typical parameters, meaning the limit cycle is born small and grows continuously as $I$ increases past $I_{Hopf}$ — the relaxation-oscillation regime 02.12.14.

Exercise 9 (hard, short-answer with rubric).

Patch-clamp recording of a single voltage-gated sodium channel shows openings of stochastic duration: at a clamped depolarisation, the channel opens and closes repeatedly, with open-times distributed approximately as an exponential with mean $⟨ τ_{open} ⟩ \approx 0.3$ ms. Explain (a) why the open-time distribution is exponential, (b) how the macroscopic conductance $g_{Na} (V_{m}, t)$ in Hodgkin-Huxley emerges from this stochastic single-channel behaviour, and (c) what the patch-clamp result adds to the HH picture.

Hint

(a) Markov property. (b) Law of large numbers across the channel population. (c) Stochasticity at low channel counts and individual channel-state structure.

Answer

(a) A two-state channel (open ↔ closed) under voltage clamp obeys first-order kinetics, $d P_{open} / d t = α (V) (1 - P_{open}) - β (V) P_{open}$ , equivalent at the single-channel level to a Markov process with constant transition rates while $V$ is held. The time spent in the open state before the next transition is therefore exponentially distributed with mean $1/ β (V)$ . The Markov property — memorylessness of state dwell — is what produces the exponential.

(b) The HH macroscopic conductance is $g_{Na} (V_{m}, t) = \overset{g}{ˉ}_{Na} m^{3} h \cdot N \cdot γ$ , where $N$ is the channel density per area and $γ$ is the single-channel conductance when open; $m^{3} h$ is the population-average open probability. By the law of large numbers, when $N$ is large (thousands of channels per square micron in squid axon membrane), the macroscopic conductance is well approximated by its mean, and the HH deterministic ODE is justified as the large- $N$ limit. At low channel counts (e.g., in fine dendrites with few channels per compartment), stochastic deviations become important, and one moves to stochastic-channel models (Fox 1997, Goldwyn-Shea-Brown 2011).

(c) Patch clamp adds: explicit channel-state structure beyond the $m^{3} h$ phenomenology (single-channel conductance levels reveal subconductance states; open-close kinetics reveal that the closed state is composed of multiple sub-states with different transition rates — a Markov chain with $\geq 4$ states is typically required to fit the data); direct measurement of channel density per area and individual-channel conductance $γ$ ; and the demonstration that channels are real, discrete molecular entities, not a continuum. Modern conductance-based models replace HH's $m^{3} h$ with full Markov chains on ${C_{1}, C_{2}, C_{3}, O, I}$ states whose transition rates are fitted to single-channel data.

Exercise 10 (hard, symbolic).

Derive the cable equation for passive spread of voltage along a cylindrical axon. Let $r_{i}$ be the axial resistivity per unit length, $r_{m}$ the transverse membrane resistance per unit length, and $c_{m}$ the capacitance per unit length. Show that

τ_{m} \partial_{t} V = λ^{2} \partial_{xx} V - V,

with $τ_{m} := r_{m} c_{m}$ the membrane time constant and $λ := r_{m} / r_{i}$ the membrane length constant.

Hint

Apply Kirchhoff's current law to a small axial element of axon of length $d x$ . The axial current obeys Ohm's law $I_{ax} (x) = - (1/ r_{i}) \partial_{x} V$ ; the transverse current per unit length is the membrane current $V / r_{m} + c_{m} \partial_{t} V$ .

Answer

Conservation of charge: the axial current entering an element at $x$ , minus the axial current leaving at $x + d x$ , equals the transverse (membrane) current leaving through the element's lateral surface:

- \frac{\partial I _{ax}}{\partial x} d x = (\frac{V}{r _{m}} + c_{m} \frac{\partial V}{\partial t}) d x .

Substitute $I_{ax} = - (1/ r_{i}) \partial_{x} V$ to get $(1/ r_{i}) \partial_{xx} V = V / r_{m} + c_{m} \partial_{t} V$ . Multiply by $r_{m}$ :

\frac{r _{m}}{r _{i}} \partial_{xx} V = V + r_{m} c_{m} \partial_{t} V, i.e., τ_{m} \partial_{t} V = λ^{2} \partial_{xx} V - V .

This is a parabolic PDE on the one-dimensional axon. The length constant $λ$ measures how far a steady-state subthreshold input spreads before decaying by $1/ e$ ; the time constant $τ_{m}$ measures how long the membrane takes to charge to a steady value. For squid axon, $λ \approx 5$ mm, $τ_{m} \approx 2$ ms. Adding the HH ionic currents on the right-hand side gives the full active cable equation that supports propagating action potentials with conduction velocity $\propto λ / τ_{m}$ — the formula underlying why thicker axons (larger $λ$ ) and myelin (smaller effective $c_{m}$ on insulated segments) both speed conduction.

Lean formalization [Intermediate+]

Mathlib does not yet cover the Hodgkin-Huxley system, the Nernst equation as a stat-mech derivation, the cable equation as a parabolic PDE on an axon, or Markov-state channel models. The closest layers are:

Mathlib.Analysis.ODE.PicardLindelof: existence and uniqueness for ODEs satisfying a Lipschitz condition.
Mathlib.Analysis.ODE.Gronwall: the Grönwall inequality.
Mathlib.MeasureTheory.MeasurableSpace.Basic and the Markov-chain infrastructure: provide the formal foundation a future single-channel patch-clamp model would build on.

There is no Mathlib definition of "ionic current across a membrane", "Hodgkin-Huxley four-variable system", or "Nernst equation". The formalisation pathway from the existing ODE layer through to a verified HH spike is laid out in lean_mathlib_gap in the frontmatter; the load-bearing primary gap is the Nernst derivation, which depends on stat-mech infrastructure that Mathlib has only in fragments.

lean_status: none reflects this. Tyler's review attests intermediate-tier correctness pending external biophysics reviewer recruitment per BIOLOGY_PLAN.md §7.

Full Hodgkin-Huxley analysis: the spike as a limit cycle in $R^{4}$ [Master]

The Hodgkin-Huxley system $(V_{m}, m, h, n)$ defines a smooth flow on $R \times [0, 1]^{3}$ governed by the four-dimensional autonomous ODE

C_{m} \dot{V}_{m} \overset{m}{˙} \dot{h} \overset{n}{˙} = - \overset{g}{ˉ}_{Na} m^{3} h (V_{m} - E_{Na}) - \overset{g}{ˉ}_{K} n^{4} (V_{m} - E_{K}) - \overset{g}{ˉ}_{L} (V_{m} - E_{L}) + I_{ext}, = α_{m} (V_{m}) (1 - m) - β_{m} (V_{m}) m, = α_{h} (V_{m}) (1 - h) - β_{h} (V_{m}) h, = α_{n} (V_{m}) (1 - n) - β_{n} (V_{m}) n,

with empirical rate functions $α_{x} (V), β_{x} (V)$ that Hodgkin and Huxley fit from voltage-clamp data on the squid giant axon at $T = 6.3°$ C ^{[Hodgkin & Huxley 1952]}. The phase space 02.12.01 is bounded ( $V_{m}$ stays in a finite range under any biophysical operating regime; $m, h, n \in [0, 1]$ are confined by the kinetics themselves), and the flow is smooth, so the Picard-Lindelöf theorem guarantees existence and uniqueness of trajectories from any initial condition.

At constant external current $I_{ext}$ , the four-variable system has either (i) a unique stable equilibrium (the resting state, for low $I_{ext}$ ); (ii) a stable limit cycle (the repetitive-spiking regime, for $I_{ext}$ above a critical value); or (iii) bistable coexistence of both attractors over a narrow window. The classical squid-axon parameters place the transition from (i) to (ii) at $I_{ext} \approx 6.3$ μA/cm² for the standard Hodgkin-Huxley constants.

The transition is a Hopf bifurcation 02.12.17 in the four-dimensional system. As $I_{ext}$ crosses the critical value $I_{H}$ , a pair of complex-conjugate eigenvalues of the Jacobian at the equilibrium crosses the imaginary axis from left to right; the equilibrium loses stability and a limit cycle is born. The bifurcation is subcritical for the classical HH parameters (the limit cycle exists just below $I_{H}$ and is initially unstable, becoming stable through a fold-of-cycles bifurcation slightly below $I_{H}$ ), which is why the spike onset is sharp and there is a narrow bistability window: small noise can flip the cell between the silent and the spiking states near threshold. Izhikevich (2007) classifies excitable systems into Type I (saddle-node-on-invariant-circle bifurcation, smooth firing-rate onset from zero) and Type II (Hopf bifurcation, firing rate onset at a finite minimum); the HH squid-axon model is Type II in this classification, and many cortical neurons are Type I.

The four-dimensional flow inherits its richness from the time-scale separation between $V_{m}$ / $m$ (fast, milliseconds) and $h$ / $n$ (slower, several milliseconds). Reducing to the slow subsystem by averaging the fast dynamics gives a two-dimensional slow flow on $(h, n)$ that captures the spike-train modulation; reducing instead by adiabatic elimination of $m$ (slaving $m$ to its instantaneous steady-state $m_{\infty} (V_{m})$ ) and combining $h$ with $n$ into a single recovery variable produces the FitzHugh-Nagumo system ^{[FitzHugh 1961]}

\dot{V} = V - V^{3} /3 - w + I, \overset{w}{˙} = ϵ (V + a - b w),

a planar caricature of HH that is geometrically tractable while keeping the essential features: a cubic fast nullcline with a region of negative differential conductance (the regenerative-excitability mechanism), a linear slow nullcline whose intersection with the cubic determines the equilibrium and the bifurcation, and an $ϵ ≪ 1$ time-scale separation that makes the limit cycle a relaxation oscillation — fast jumps between the outer branches of the cubic punctuated by slow transits along them. The relaxation-oscillation analysis is precisely the Liénard / Van der Pol theory 02.12.14 adapted to the FN-form nonlinearity, with the Hopf bifurcation in $I$ playing the role of the bifurcation in the Van der Pol $μ$ parameter. The Morris-Lecar model (Morris-Lecar 1981 Biophys. J. 35, 193-213) is a planar HH-like model for muscle that admits both Type I (saddle-node-on-invariant-circle) and Type II (Hopf) excitability regimes depending on the calcium-activation steepness, and is the cleanest pedagogical setting in which to see the bifurcation classification.

Single-channel biophysics from patch-clamp [Master]

The voltage-clamp data of Hodgkin and Huxley measured the macroscopic membrane current through populations of millions of channels; the 1976 patch-clamp recordings of Neher and Sakmann ^{[Neher & Sakmann 1976]} resolved the individual channel as a discrete molecular gate flipping open and shut. A single voltage-gated sodium channel at a clamped depolarisation opens for a few hundred microseconds at a time, then closes again, with a single-channel conductance $γ \approx 10$ - $20$ pS that is the same from opening to opening and the same across channels. The macroscopic conductance is the channel density times the open probability times $γ$ :

g_{X} (V_{m}, t) = N_{X} \cdot γ_{X} \cdot P_{open}^{X} (V_{m}, t) .

For HH, $N_{X} γ_{X} = \overset{g}{ˉ}_{X}$ and $P_{open}^{X} = m^{3} h$ or $n^{4}$ . The single-channel measurement directly resolves $γ$ and (by counting channel openings) $N$ , separating the two factors that the macroscopic conductance lumps together.

Open-time and closed-time distributions of patch-clamp records show that the gating dynamics is governed by a continuous-time Markov chain on a finite state space — typically four or five states for the sodium channel: three or four closed states $C_{1} \to C_{2} \to C_{3} \to O$ with state-dependent voltage-dependent rate constants, plus an inactivated state $I$ accessible only from the open state and recovering slowly. The Hodgkin-Huxley $m^{3} h$ form is approximately reproduced by this Markov chain in the limit where the three closed-to-open transitions have the same voltage dependence, and where the inactivation gate operates independently of activation (the factorisation hypothesis). High-resolution single-channel data show systematic deviations from $m^{3} h$ : closed-state durations are not exponentially distributed (multi-state structure); inactivation depends on the open-state occupancy (a non-factorisable form); and recovery from inactivation goes through multiple intermediate states. The modern conductance-based model is Markov-chain-based throughout (Patlak 1991 Physiol. Rev. 71, 1047-1080; Vandenberg-Bezanilla 1991 Biophys. J. 60, 1511-1533).

The connection to the macroscopic deterministic flow runs through the law of large numbers for finite-state continuous-time Markov chains: the fraction of channels in each state converges, as the channel count grows, to the deterministic occupancy obeying the corresponding deterministic ODE. At physiological channel densities (thousands per square micron in squid axon) the deterministic limit is excellent and Hodgkin-Huxley is well-justified; at low densities (sparse dendritic distributions, single-vesicle release at synapses) stochastic effects matter and conductance-based stochastic simulation algorithms (Gillespie-style for the chain transitions) become necessary (Skaugen-Walløe 1979 Acta Physiol. Scand. 107, 343-363; Fox 1997 Biophys. J. 72, 2068-2074; Goldwyn-Shea-Brown 2011 PLoS Comput. Biol. 7, e1002247).

Active propagation along the axon: the cable equation with HH currents [Master]

Subthreshold spread of voltage along an unmyelinated axon obeys the passive cable equation $τ_{m} \partial_{t} V = λ^{2} \partial_{xx} V - V$ , with membrane time constant $τ_{m}$ and length constant $λ$ , derived from Kirchhoff's current law on a cylindrical conductor. The equation is parabolic; signals diffuse, decay exponentially with distance over a scale $λ$ , and the membrane acts as a low-pass filter. For squid axon ( $r_{a} \approx 30 Ω \cdot cm$ , $r_{m} \approx 2000 Ω \cdot cm^{2}$ , $c_{m} \approx 1 μ F/cm^{2}$ , radius $a = 250$ μm) the passive length constant is $λ = a r_{m} /2 r_{a} \approx 6.5$ mm and the time constant $τ_{m} = r_{m} c_{m} \approx 2$ ms.

Adding the HH ionic currents on the right-hand side gives the active cable equation

\frac{a}{2 r _{a}} \partial_{xx} V = C_{m} \partial_{t} V + I_{ion} (V, m, h, n),

a reaction-diffusion PDE in which the diffusion term spreads voltage passively along the axon and the local reaction term regenerates the spike at each location. Hodgkin and Huxley used this equation, with the HH ionic currents, to predict the propagation velocity of the squid action potential from voltage-clamp data alone — about $21$ m/s — in agreement with the directly measured value, without fitting any free parameter to the propagation experiment ^{[Hodgkin & Huxley 1952]}. The agreement is one of the canonical predictive successes in twentieth-century biology.

Myelinated axons replace the continuous active membrane with discrete nodes of Ranvier (active patches, with high channel density) separated by myelinated internodes (passive insulating sheaths, with low capacitance per unit length and high transverse resistance). The voltage at one node spreads passively along the internode — fast and almost lossless — and triggers regeneration at the next node. The conduction velocity in a myelinated axon scales linearly with diameter ( $v \propto d$ ) rather than as the square root ( $v \propto d$ ) of an unmyelinated one, the geometric origin of the fifty-fold speedup at fixed axon diameter. Back-propagation of action potentials into the dendritic tree (Stuart-Sakmann 1994 Nature 367, 69-72) couples the spike output back to dendritic computation and is a central element of synaptic plasticity in cortical pyramidal neurons.

Calcium and the link to plasticity [Master]

The action potential's last act, in many neurons, is to depolarise voltage-gated calcium channels — distinct molecular machines from the sodium channels of the upstroke — and admit a brief calcium current. Intracellular calcium acts as a second messenger: it activates calcium-dependent potassium channels (driving the afterhyperpolarisation), triggers neurotransmitter release at presynaptic terminals (the link to synaptic transmission 17.07.04 pending), and engages calcium-dependent signalling cascades that, on longer timescales, alter the strength of individual synapses. The action potential is therefore not only an information-bearing pulse but also a calcium-delivery event whose calcium load encodes information about firing rate and recent firing history, with consequences for learning and memory through long-term potentiation and depression at the synapses contacted by the spiking neuron.

Modern conductance-based models embed HH-style sodium and potassium currents in a richer ionic-current ensemble: low- and high-voltage-activated calcium currents, calcium-dependent potassium currents (BK and SK), HCN (hyperpolarisation-activated cation) currents, A-type potassium, and others — each with its own gating variables and voltage / calcium dependence. The dynamical-systems machinery developed for HH extends naturally to these higher-dimensional flows, and the bifurcation classification (Type I vs Type II, regular spiking vs bursting, the various codimension-two bifurcations on parameter planes) is the organising scheme for understanding how the diverse firing patterns of real neurons are generated.

Connections [Master]

Nernst equation and the canonical ensemble 11.04.01 pending. The action-potential unit is the densest biological application of the canonical-ensemble Boltzmann factor: every equilibrium potential is a stat-mech equilibrium calculation, and the GHK extension is the multi-species generalisation. The reciprocal hook from 11.04.01 pending back to here is proposed; this unit consumes the canonical-ensemble result as a load-bearing primitive.
Maxwell's equations 10.04.01 pending. The membrane is a capacitor; the ionic currents are charge transport; the propagating spike is a current pulse with an associated extracellular field. Treating the action potential as a problem in classical electromagnetism on a cellular geometry is the route by which Maxwell's equations enter cellular biology. The cable equation is the relevant simplification — quasi-static (no radiation), one-dimensional (cylindrical symmetry of the axon), with the membrane providing the boundary condition.
ODE phase space 02.12.01. The HH four-variable system is one of the most-studied examples of an autonomous flow on a four-dimensional phase space; every classical statement about flows, vector fields, and integral curves applies. The spike is a limit cycle, the resting state is a stable equilibrium, the threshold is a separatrix (in the planar reduction).
Limit cycles and Liénard / Van der Pol 02.12.14. The reduction of HH to FitzHugh-Nagumo produces a Liénard-type planar system whose limit cycle is the spike. Relaxation-oscillation asymptotics in the singular limit $ϵ \to 0$ give analytic expressions for the period and the spike shape.
Bifurcation theory 02.12.17. The transition from silence to repetitive firing as $I_{ext}$ varies is a codimension-one bifurcation — supercritical or subcritical Hopf in HH and most Type II neurons; saddle-node-on-invariant-circle in Type I. The bifurcation classification organises the diverse firing patterns of biological neurons.
Ion-channel chemistry 15.13.02 pending (pending — chem §15 not yet built). Each gating variable of HH is a phenomenological summary of the channel's molecular conformational dynamics; the chem-side unit on ion-channel structure and the selectivity-filter coordination chemistry is the level-below substrate.
Membrane structure and transport [17.02] (pending). The lipid bilayer's impermeability to ions and the role of integral membrane proteins as the only paths through is the structural prerequisite for everything in this unit.
Cell signalling and calcium 17.07.04 pending (pending). The action potential's calcium load drives second-messenger cascades that connect membrane excitability to gene expression and synaptic plasticity.
Systems neuroscience 18.05.02 pending (pending — bio §18 not yet built). The action potential is the atomic unit of nervous-system signalling on which the integrated-physiology side builds synaptic transmission, neural circuits, and behaviour.
Phil of mind on the neural substrate 20.06.01 pending (pending — phil §20 tiered units not yet introduced). The cellular-molecular foundation of cognition: whether neural computation reduces to action-potential trains is the substrate-side of the question phil-of-mind formulates abstractly.
Translation 17.05.03 pending. Voltage-gated ion channels are membrane proteins produced by translation; their density at the axon membrane depends on translation rates and trafficking efficiency. Conversely, action potentials consume ATP (for Na+/K+ pump activity), and the energy budget of a neuron is dominated by the cost of translation to replace ion channels and synaptic proteins. The action potential and translation are thus linked through the cell's energy budget: each spike has a downstream translational cost.

Historical & philosophical context [Master]

The action potential's modern theory comes out of a roughly seventy-year experimental program from Bernstein's 1902 Untersuchungen zur Thermodynamik der bioelektrischen Ströme ^{[Bernstein 1902]} — which proposed that nerve membranes are selectively permeable to potassium at rest and lose that selectivity during activity — through to the molecular cloning of the sodium channel in 1984 (Noda et al. Nature 312, 121-127) and the structural biology of voltage-gated channels at angstrom resolution from the early 2000s onward (the MacKinnon laboratory's crystal structures of bacterial voltage-gated potassium channels — KcsA in 1998, KvAP in 2003 — earned MacKinnon the 2003 Nobel Prize in chemistry).

The decisive midcentury step was the Hodgkin-Huxley voltage-clamp program at Plymouth Marine Laboratory and Cambridge from 1939 (interrupted by the war) through 1952 ^{[Hodgkin & Huxley 1952]}. With Bernard Katz, they developed the voltage clamp on the squid giant axon (the squid giant axon being usable as an experimental system was itself a discovery, by Young in 1936), separated the sodium and potassium currents by ionic substitution and pharmacological blockade, fit the empirical rate functions of $m, h, n$ from voltage-clamp records at various clamped potentials, and showed that the resulting four-variable ODE system predicts the spike shape, threshold, refractory period, and propagation velocity quantitatively. The 1952 series (five papers in Journal of Physiology volumes 116 and 117) is one of the cleanest predictive successes in twentieth-century biology: a model fit to one experiment (voltage-clamp current-voltage relations) predicted, with no free parameters, the outcome of an independent experiment (propagation velocity of an action potential along an unclamped axon). Hodgkin, Huxley, and Eccles shared the 1963 Nobel Prize in physiology or medicine.

The patch-clamp technique developed by Neher and Sakmann in Göttingen from the mid-1970s onward ^{[Neher & Sakmann 1976]} resolved single-channel openings as discrete electrical events, opening the molecular and statistical biology of channel gating as a tractable experimental field. Neher and Sakmann shared the 1991 Nobel Prize in physiology or medicine. The structural biology of channels in the 2000s (MacKinnon's KcsA, KvAP, and subsequent voltage-gated sodium-channel structures from Catterall and others) closed the loop between Hodgkin and Huxley's empirical kinetic constants and the atomistic motion of the channel protein: the S4 voltage sensor's charged arginine residues move outward across the membrane under depolarisation, mechanically opening a hydrophobic gate at the cytoplasmic end of the pore.

The Hodgkin-Huxley framework is more than a model of a specific cell. It is the canonical example of how empirical kinetic constants fit at the macroscopic scale predict the dynamics of an excitable cellular system, the cellular-biology analog of phenomenological transport coefficients in stat-mech-of-fluids. The success of the framework — its quantitative agreement with the experimental spike from voltage-clamp data alone — is also a benchmark against which all later mechanistic models of cellular dynamics are calibrated. The price of the success is that the model is phenomenological: $m, h, n$ are not physical positions of any specific molecular gate; they are dimensionless population averages with the structural origin only inferred. Patch-clamp and structural biology have since identified the molecular meanings, vindicating the inference in broad outline while showing where the four-variable phenomenology breaks down in detail.

Bibliography [Master]

Primary literature.

Hodgkin, A. L. & Huxley, A. F., "A quantitative description of membrane current and its application to conduction and excitation in nerve", J. Physiol. 117 (1952), 500–544.

Hodgkin, A. L., Huxley, A. F. & Katz, B., "Measurement of current-voltage relations in the membrane of the giant axon of Loligo", J. Physiol. 116 (1952), 424–448.

Hodgkin, A. L. & Katz, B., "The effect of sodium ions on the electrical activity of the giant axon of the squid", J. Physiol. 108 (1949), 37–77.

Neher, E. & Sakmann, B., "Single-channel currents recorded from membrane of denervated frog muscle fibres", Nature 260 (1976), 799–802.

Goldman, D. E., "Potential, impedance, and rectification in membranes", J. Gen. Physiol. 27 (1943), 37–60.

Nernst, W., "Die elektromotorische Wirksamkeit der Ionen", Z. Phys. Chem. 4 (1889), 129–181.

Bernstein, J., Untersuchungen zur Thermodynamik der bioelektrischen Ströme, Pflügers Archiv (1902), 173–207.

FitzHugh, R., "Impulses and physiological states in theoretical models of nerve membrane", Biophys. J. 1 (1961), 445–466.

Nagumo, J., Arimoto, S. & Yoshizawa, S., "An active pulse transmission line simulating nerve axon", Proc. IRE 50 (1962), 2061–2070.

Morris, C. & Lecar, H., "Voltage oscillations in the barnacle giant muscle fiber", Biophys. J. 35 (1981), 193–213.

Noda, M. et al., "Primary structure of Electrophorus electricus sodium channel deduced from cDNA sequence", Nature 312 (1984), 121–127.

Doyle, D. A. et al., "The structure of the potassium channel: molecular basis of K+ conduction and selectivity", Science 280 (1998), 69–77.

Jiang, Y. et al., "X-ray structure of a voltage-dependent K+ channel", Nature 423 (2003), 33–41.

Stuart, G. J. & Sakmann, B., "Active propagation of somatic action potentials into neocortical pyramidal cell dendrites", Nature 367 (1994), 69–72.

Textbook and monograph.

Hille, B., Ion Channels of Excitable Membranes, 3rd ed. (Sinauer, 2001).

Kandel, E. R., Schwartz, J. H., Jessell, T. M., Siegelbaum, S. A. & Hudspeth, A. J., Principles of Neural Science, 5th ed. (McGraw-Hill, 2013).

Alberts, B. et al., Molecular Biology of the Cell, 6th ed. (Garland, 2014).

Boron, W. F. & Boulpaep, E. L., Medical Physiology, 3rd ed. (Elsevier, 2017).

Koch, C., Biophysics of Computation: Information Processing in Single Neurons (Oxford University Press, 1999).

Sakmann, B. & Neher, E. (eds), Single-Channel Recording, 2nd ed. (Plenum, 1995).

Izhikevich, E. M., Dynamical Systems in Neuroscience: The Geometry of Excitability and Bursting (MIT Press, 2007).

Dayan, P. & Abbott, L. F., Theoretical Neuroscience: Computational and Mathematical Modeling of Neural Systems (MIT Press, 2001).

Tuckwell, H. C., Introduction to Theoretical Neurobiology, vols. 1-2 (Cambridge University Press, 1988).

Wave 1 biology seed unit, produced manually 2026-05-18 (per docs/plans/BIOLOGY_PLAN.md §6 — densest cross-domain prereq node in any of the five domains; manual production with human-author judgement on cross-cite choices). Status remains draft pending Tyler's review, external mol-cell-bio + computational-neuroscience reviewer recruitment, and the §11 next-actions retro per BIOLOGY_PLAN. The cross-domain prereqs to chem §15 (ion-channel structure) and bio §17.02 (membrane transport) are pending pending-edge registration in manifests/deps.json by the integrator pass; the physics prereqs $[10.04.01]$ and $[11.04.01]$ exist as draft units pending their own shipping, and the link contract is stress-tested in real conditions per BIOLOGY_PLAN §6.1.

Prerequisites

02.12.01
02.12.14

Tier anchors

beginner: Crash Course Biology / Anatomy & Physiology (neuron episodes); Nick Lane, *The Vital Question* (W. W. Norton 2015), chapters on the proton-motive force and the energetic basis of excitability
intermediate: Alberts et al., *Molecular Biology of the Cell* (6th ed., Garland 2014), Ch. 11 *Membrane Transport of Small Molecules and the Electrical Properties of Membranes*; Boron & Boulpaep, *Medical Physiology* (3rd ed., Elsevier 2017), Ch. 7 *Electrophysiology of the Cell Membrane*; Kandel, Schwartz, Jessell, Siegelbaum, Hudspeth, *Principles of Neural Science* (5th ed., McGraw-Hill 2013), Part II *Cell and Molecular Biology of the Neuron*
master: Hille, *Ion Channels of Excitable Membranes* (3rd ed., Sinauer 2001) — the canonical reference; Koch, *Biophysics of Computation* (Oxford University Press 1999), Chs. 6-7; Kandel et al., *Principles of Neural Science* (5th ed.) — advanced sections including patch-clamp and Markov-state modelling; Hodgkin & Huxley 1952 (*J. Physiol.* 117, 500-544) — the four foundational papers culminating in the quantitative description; Neher & Sakmann 1976 (*Nature* 260, 799-802) — patch-clamp recording of single-channel currents; Sakmann & Neher 1995 *Single-Channel Recording* (2nd ed., Plenum); Izhikevich, *Dynamical Systems in Neuroscience* (MIT Press 2007) — bifurcation analysis of HH and reduced models

References

TODO_REF pending
Hodgkin & Huxley 1952 — A quantitative description of membrane current and its application to conduction and excitation in nerve · J. Physiol. 117 (1952) 500-544; the fourth and culminating paper of the 1952 series in which the four-variable model with gating variables $m$, $h$, $n$ is fit to squid-giant-axon voltage-clamp data and used to predict the spike shape and conduction velocity from first principles; Nobel Prize 1963 · see docs/catalogs/NEED_TO_SOURCE.md#bio-wave1-hh1952
TODO_REF pending
Hodgkin, Huxley & Katz 1952 — Measurement of current-voltage relations in the membrane of the giant axon of Loligo · J. Physiol. 116 (1952) 424-448; the voltage-clamp methodology paper that made the 1952 series possible · see docs/catalogs/NEED_TO_SOURCE.md#bio-wave1-hhk1952
TODO_REF pending
Neher & Sakmann 1976 — Single-channel currents recorded from membrane of denervated frog muscle fibres · Nature 260 (1976) 799-802; first patch-clamp recording of single ion-channel openings; Nobel Prize 1991 · see docs/catalogs/NEED_TO_SOURCE.md#bio-wave1-ns1976
TODO_REF pending
Hille — Ion Channels of Excitable Membranes (3rd ed., Sinauer 2001) · Chs. 1-4 (overview, classical biophysics, selectivity); Chs. 5-7 (voltage-gated Na+, K+, Ca2+ channels); Ch. 18 (modelling, including HH and Markov-state extensions) · see docs/catalogs/NEED_TO_SOURCE.md#bio-wave1-hille-ionchannels
TODO_REF pending
Alberts et al. — Molecular Biology of the Cell (6th ed., Garland 2014) · Ch. 11 — Membrane Transport of Small Molecules and the Electrical Properties of Membranes; §§ Ion Channels and the Electrical Properties of Membranes (pp. 615-635) · see docs/catalogs/NEED_TO_SOURCE.md#bio-wave1-mboc-ch11
TODO_REF pending
Kandel, Schwartz, Jessell, Siegelbaum & Hudspeth — Principles of Neural Science (5th ed., McGraw-Hill 2013) · Ch. 7 *Membrane Potential and the Passive Electrical Properties of the Neuron*; Ch. 8 *Local Signalling: Passive Electrical Properties of the Neuron*; Ch. 9 *Propagated Signalling: The Action Potential*; Ch. 10 *Overview of Synaptic Transmission* · see docs/catalogs/NEED_TO_SOURCE.md#bio-wave1-kandel-pns
TODO_REF pending
Koch — Biophysics of Computation: Information Processing in Single Neurons (Oxford University Press 1999) · Chs. 6-7 (Hodgkin-Huxley model and reduced models including FitzHugh-Nagumo and Morris-Lecar); Ch. 11 (cable theory and dendritic integration) · see docs/catalogs/NEED_TO_SOURCE.md#bio-wave1-koch-biophys
TODO_REF pending
Boron & Boulpaep — Medical Physiology (3rd ed., Elsevier 2017) · Ch. 6 *Electrophysiology of the Cell Membrane*; Ch. 7 *Electrical Excitability and Action Potentials*; the intermediate-tier integrated-physiology reference · see docs/catalogs/NEED_TO_SOURCE.md#bio-wave1-boron-boulpaep
TODO_REF pending
Izhikevich — Dynamical Systems in Neuroscience: The Geometry of Excitability and Bursting (MIT Press 2007) · Ch. 5 (conductance-based models, HH analysis); Ch. 6 (bifurcation analysis, Hopf vs saddle-node-on-invariant-circle classifications of spike onset); Ch. 8 (reduction to planar models including FitzHugh-Nagumo) · see docs/catalogs/NEED_TO_SOURCE.md#bio-wave1-izhikevich
TODO_REF pending
Nernst, W. — Zur Kinetik der in Lösung befindlichen Körper. Erste Abhandlung: Theorie der Diffusion · Z. Phys. Chem. 2 (1888) 613-637; the originator paper for what becomes the Nernst equation through Nernst (1889) Z. Phys. Chem. 4, 129-181 *Die elektromotorische Wirksamkeit der Ionen* — the equilibrium potential of an electrolyte separated by a selectively permeable barrier · see docs/catalogs/NEED_TO_SOURCE.md#bio-wave1-nernst1888
TODO_REF pending
Goldman, D. E. — Potential, impedance, and rectification in membranes · J. Gen. Physiol. 27 (1943) 37-60; the constant-field assumption and the multi-ion GHK equation that generalises Nernst when the membrane is permeable to several ions at once · see docs/catalogs/NEED_TO_SOURCE.md#bio-wave1-goldman1943
TODO_REF pending
Hodgkin, A. L. & Katz, B. — The effect of sodium ions on the electrical activity of the giant axon of the squid · J. Physiol. 108 (1949) 37-77; first quantitative confirmation that Na+ entry drives the spike upstroke and that the peak voltage tracks the Na+ Nernst potential; the GHK formula appears as the resting-potential equation · see docs/catalogs/NEED_TO_SOURCE.md#bio-wave1-hk1949
TODO_REF pending
FitzHugh, R. — Impulses and physiological states in theoretical models of nerve membrane · Biophys. J. 1 (1961) 445-466; the planar reduction of HH that retains the spike and threshold structure with two variables (later refined by Nagumo, Arimoto & Yoshizawa 1962 Proc. IRE 50, 2061-2070 with the tunnel-diode realisation) · see docs/catalogs/NEED_TO_SOURCE.md#bio-wave1-fitzhugh1961
tong
raw/pdfs/statphys/statphys.pdf · §1.3 The Canonical Ensemble — the partition-function and Boltzmann-factor foundation that the Nernst equation pulls from; §2 Classical Gases for ideal-gas thermodynamic background

Reviewer

Tyler (pending external molecular/cellular biologist + computational neuroscientist reviewers per BIOLOGY_PLAN §7 — highest priority dual recruit; the cellular-neuroscience specialty is the single highest-leverage hire for the §17 axis, and the computational-neuroscience reviewer is needed to attest the HH dynamical-systems content at Master tier)

Estimated time

beginner: 18m
intermediate: 35m
master: 65m

Intuition [Beginner]

Visual [Beginner]

Worked example [Beginner]

Check your understanding [Beginner]

Formal definition [Intermediate+]

Counterexamples to common slips

Key theorem with proof [Intermediate+]

Worked example at intermediate level: the resting potential from GHK

Exercises [Intermediate+]

Lean formalization [Intermediate+]

Full Hodgkin-Huxley analysis: the spike as a limit cycle in R4 [Master]

Single-channel biophysics from patch-clamp [Master]

Active propagation along the axon: the cable equation with HH currents [Master]

Calcium and the link to plasticity [Master]

Connections [Master]

Historical & philosophical context [Master]

Bibliography [Master]

Full Hodgkin-Huxley analysis: the spike as a limit cycle in $R^{4}$ [Master]