Physiology
Action Potential
~1 ms voltage spike from -70 to +40 mV propagates along axons at up to 120 m/s — voltage-gated Na+/K+ channels
An action potential is a brief, all-or-nothing voltage spike that neurons use to send signals down their axons. The neuronal membrane normally sits at -70 mV; when a stimulus depolarizes it past about -55 mV, voltage-gated Na+ channels snap open in under 0.5 ms and the inside of the cell swings up to about +40 mV. Then Na+ channels inactivate and voltage-gated K+ channels open, dragging the voltage back down. A roughly 2 ms refractory period follows, during which the patch cannot fire again — which is what forces the spike to travel one direction down the axon. Conduction speeds range from 0.5 m/s in thin unmyelinated fibers to 120 m/s in heavily myelinated ones, where the spike "jumps" between nodes of Ranvier in saltatory conduction. Hodgkin and Huxley worked out the whole story on the squid giant axon in 1952 and shared the 1963 Nobel Prize.
- Resting potential-70 mV (typical neuron)
- Peak voltage~+40 mV
- Rising phase~0.5 ms (Na+ influx)
- Refractory~2 ms absolute
- Conduction speed0.5–120 m/s
- Solved byHodgkin & Huxley 1952 (Nobel 1963)
Interactive visualization
Press play, or step through manually. The visualization is yours to drive — try it before reading on.
Watch the 60-second explainer
A condensed visual walkthrough — narrated, captioned, under a minute.
Why action potentials matter
- Universal currency of nervous-system communication. Every neuron in every animal that has nerves uses action potentials. From a 1 mm C. elegans with 302 neurons to a human cortex with about 86 billion, the spike is the message. Information is encoded in spike timing and rate — a single neuron can fire from 0.1 Hz to ~1000 Hz.
- Speed matters for survival. A garden snail's ~0.5 m/s reflex is fine; a cheetah's pounce or a human's pain withdrawal needs 80–120 m/s motor output. Vertebrate evolution of myelin, repeatedly across taxa, was driven by the need for faster long-distance signaling without making axons impractically thick.
- Drug targets — local anesthetics, antiepileptics. Lidocaine blocks voltage-gated Na+ channels in their open state, preventing pain neurons from firing. Carbamazepine and phenytoin similarly stabilize the inactivated state and treat epilepsy. Tetrodotoxin (TTX) from pufferfish is the highly specific Na+ channel blocker used to dissect the currents experimentally — and is lethal at 1 mg in humans.
- Hodgkin-Huxley model is foundational quantitative biology. The four-equation system fit to squid giant axon data in 1952 was one of the first cases of computational biology. Modern neural simulation software (NEURON, NEST, Brian) still uses Hodgkin-Huxley-style ODEs at its core; the equations work for any voltage-gated channel given the right gating parameters.
- Direct mechanism for clinical disorders. Multiple sclerosis demyelinates CNS axons, slowing or blocking conduction. Guillain-Barré demyelinates peripheral axons. Mutations in voltage-gated Na+ channels cause genetic epilepsies, long-QT syndrome (cardiac SCN5A), and inherited erythromelalgia (SCN9A pain).
- Cardiac variant differs by 200x in duration. Cardiac ventricular action potentials last about 200–400 ms (versus 1 ms in neurons) because of an L-type Ca2+ channel plateau. The plateau prevents tetanus in heart muscle and is targeted by class III antiarrhythmics like amiodarone that prolong it.
- Energy budget anchor. Restoring ion gradients after spiking consumes about 3.5 ATP per spike per micrometer of axon. Brain action-potential traffic accounts for roughly 20 percent of human resting metabolic rate despite the brain being only 2 percent of body mass.
Common misconceptions
- Na+/K+ ATPase pumps drive the spike. No — the spike is driven by passive ion flow through voltage-gated channels along pre-existing gradients. The pump's job is to maintain the gradient between spikes, not to generate the spike itself. A spike still fires for several seconds after pumps are blocked, until gradients run down.
- Larger stimulus produces larger spike. Action potentials are all-or-nothing once threshold is crossed. Larger stimuli produce more spikes per second (higher firing rate) or recruit more neurons, not bigger spikes. Amplitude is essentially fixed at about 110 mV peak-to-trough.
- Information is in the spike shape. Spike shape is roughly stereotyped within a neuron type. The information lies in timing (precise spike times, intervals between spikes) and rate (number of spikes per unit time). Cortical neurons can encode under sub-millisecond temporal precision.
- Myelin contains the channels. Myelin is mostly inert wrapping. Voltage-gated Na+ channels are concentrated at the tiny unmyelinated nodes of Ranvier (about 1 µm wide), every 1–2 mm along the axon. The myelin between nodes acts as a passive insulator.
- Action potentials propagate by diffusion. Ions diffuse only over micrometers per millisecond — way too slow. Propagation is by local-circuit currents: depolarization at one patch flows passively along the axoplasm to the next patch and triggers regeneration there. Diffusion sets the channel kinetics, not the wave speed.
- The spike exhausts the cell's salt. Each spike moves only about 10−12 mol of Na+ across a typical axon section — a fraction so tiny that intracellular Na+ rises by less than 1 part in a thousand per spike in a small neuron. Even after thousands of spikes, the gradient is maintained.
How the action potential works
At rest, the neuronal membrane sits at about -70 mV because it is dominated by K+ leak permeability. A subthreshold depolarization (synaptic input, sensory transduction, electrical stimulus) brings the voltage closer to threshold. Crossing about -55 mV opens voltage-gated Na+ channels — these have four S4 voltage-sensing helices and an activation gate that snaps open in under 100 µs. Na+ flows in along its electrochemical gradient (intracellular ~14 mM, extracellular ~145 mM, equilibrium potential ~+60 mV) and depolarizes the membrane further, opening more Na+ channels in a regenerative positive-feedback loop. The membrane voltage rockets toward +40 mV in about 0.5 ms.
About 1 ms after activation, an inactivation gate swings into the Na+ channel pore and stops the inward current. Slightly later, voltage-gated K+ channels (delayed rectifiers) open and K+ flows out along its gradient, repolarizing the membrane. K+ channels stay open briefly after -70 mV, producing a small afterhyperpolarization to about -75 mV. Total spike duration is around 1 ms in mammalian axons. Propagation occurs because depolarization at one patch passively flows down the axon (decaying with the cable's length constant, typically 0.1–1 mm) to the next patch, where it brings that patch above threshold and triggers regeneration. The refractory region behind the wave cannot fire again, forcing unidirectional travel. In myelinated axons, regeneration happens only at nodes of Ranvier and depolarization passively jumps myelinated internodes — saltatory conduction increases speed by 5–50x for a given axon diameter.
Action potential vs graded potential
| Property | Action potential | Graded (post-synaptic / receptor) potential |
|---|---|---|
| Amplitude | All-or-nothing, ~110 mV | Continuous, sub-mV to ~30 mV |
| Duration | ~1 ms (mammalian neuron) | 5–100 ms |
| Channels | Voltage-gated Na+ and K+ | Ligand-gated or mechanically-gated |
| Propagation | Active, regenerative, no decay | Passive, decays over length constant (0.1–1 mm) |
| Threshold required? | Yes (~-55 mV) | No |
| Refractory period | ~2 ms absolute | None |
| Summation | Cannot summate (all-or-nothing) | Spatial and temporal summation |
| Information encoding | Spike timing and rate | Amplitude and time integral |
Myelinated vs unmyelinated propagation
| Property | Myelinated axon (saltatory) | Unmyelinated axon (continuous) |
|---|---|---|
| Conduction velocity | 10–120 m/s | 0.5–2 m/s |
| Sheath | Multi-layer phospholipid wrapping (oligodendrocyte / Schwann cell) | None |
| Na+ channel distribution | Concentrated at nodes of Ranvier (~1 µm gaps) | Uniform along membrane |
| Internode length | 1–2 mm typically | N/A |
| Energy per unit length | Lower — only nodes spike | Higher — full membrane spikes |
| Erlanger-Gasser class | A-alpha, A-beta (large myelinated) | C fibers (slow pain, autonomic) |
| Failure mode | Demyelination (MS, Guillain-Barré) | Sodium channel block, ion gradient loss |
| Vertebrate evolution | Convergent — appears in jawed vertebrates 360 Mya | Ancestral state across animals |
Famous experiments
- Hodgkin-Huxley 1952 squid giant axon series. Alan Hodgkin and Andrew Huxley published five papers in J. Physiol. that intracellularly recorded the action potential, invented the voltage-clamp, separated Na+ and K+ currents by ion substitution, and modeled the whole thing with four ODEs. The model predicted action potential shape and conduction velocity from first principles. Nobel Prize 1963 (shared with John Eccles).
- Cole 1939 voltage-clamp prototype. Kenneth Cole at Woods Hole built the first voltage-clamp circuit in 1939, holding the squid axon membrane at fixed voltage while recording current. Hodgkin and Huxley adopted and refined the technique on their visit to Plymouth and used it to dissect ionic currents. Cole continued working on cable theory of neural propagation.
- Erlanger-Gasser 1937 axon classification. Joseph Erlanger and Herbert Gasser used cathode ray oscilloscopes to record from peripheral nerves, identifying distinct A, B, and C fiber classes based on conduction velocity and diameter. A-alpha (motor, proprioception) at 80–120 m/s, A-delta (cold, fast pain) at 5–30 m/s, C (slow pain, autonomic) at 0.5–2 m/s. Nobel Prize 1944.
- Neher-Sakmann 1976 patch clamp. Erwin Neher and Bert Sakmann at Göttingen recorded current through individual ion channels by sealing a glass micropipette to a tiny patch of membrane. They directly observed single-channel openings of about 2 pA lasting milliseconds, confirming that the macroscopic current Hodgkin and Huxley had modeled was the sum of many discrete channel events. Nobel Prize 1991.
- Tasaki 1939 saltatory conduction. Ichiji Tasaki demonstrated experimentally that excitation in myelinated frog axons jumps from node to node rather than propagating continuously — the direct evidence for saltatory conduction. Confirmed and refined by Lillian Huang and Andrew Huxley in 1949.
Frequently asked questions
Why is the resting membrane potential -70 mV?
The resting potential reflects the balance of selective ion permeabilities and concentration gradients. Inside a typical mammalian neuron, K+ is about 140 mM and Na+ about 14 mM; outside, K+ is about 4 mM and Na+ about 145 mM. The membrane at rest is roughly 25 times more permeable to K+ than to Na+ (because K+ leak channels are open and Na+ channels are mostly closed), so the membrane potential sits close to the K+ equilibrium potential of about -90 mV calculated from the Nernst equation. A small Na+ leak nudges it up to about -70 mV. The Na+/K+ ATPase pump consumes ATP to maintain the gradient — pumping out 3 Na+ for every 2 K+ in — and contributes a few mV of hyperpolarization itself. All three effects (Nernst, leak ratio, pump) together give the canonical -70 mV resting value.
What triggers the rising phase?
Depolarization above a threshold of about -55 mV opens voltage-gated Na+ channels. Each channel has four S4 voltage sensors that detect membrane voltage and an activation gate that snaps open within microseconds. Once open, Na+ rushes in along its gradient, pushing the membrane potential toward the Na+ equilibrium potential (about +60 mV). That further depolarization opens more Na+ channels — positive feedback — so the rise is regenerative and explosive, completing in about 0.5 ms. The rising phase is what makes the action potential all-or-nothing: below threshold, the inward current is too small to recruit enough channels and the response decays; above threshold, the regenerative loop fires the full spike. This is why neurons encode information in spike timing and rate, not amplitude.
What ends the spike?
Two events terminate the depolarization. First, voltage-gated Na+ channels inactivate — a hinged inactivation gate (the IFM motif on the cytoplasmic side) swings into the pore about 1 ms after activation, plugging it. The channel cannot reopen until the membrane repolarizes, producing the absolute refractory period. Second, voltage-gated K+ channels (delayed rectifiers) open about 1 ms later than Na+ channels, allowing K+ to flow out and pull the membrane back toward its equilibrium potential. The combination of Na+ inactivation and K+ efflux produces the falling phase. K+ channels stay open briefly after the membrane has reached -70 mV, causing a small undershoot (hyperpolarization to about -75 mV) before the system returns to rest.
What is the refractory period and why does it matter?
The absolute refractory period is the ~2 ms window after a spike during which Na+ channels remain inactivated and no stimulus, however large, can produce another action potential. This is followed by a relative refractory period of a few more milliseconds in which a stronger-than-normal stimulus is needed because K+ channels are still open. The refractory period serves two crucial functions: it sets a maximum firing rate of about 500–1000 Hz and it enforces unidirectional propagation along the axon. After a spike has passed a region, the just-fired patch is refractory and cannot be re-excited by the still-active region behind it, so the wave can only move forward. This is why action potentials do not bounce back along the axon.
How does myelination speed up propagation?
Myelin is a multi-layer phospholipid wrapping produced by oligodendrocytes in the central nervous system and Schwann cells in the peripheral nervous system. It dramatically increases the membrane resistance and decreases its capacitance, so passive (electrotonic) current spreads further along the axon before decaying. Voltage-gated Na+ channels are concentrated at small unmyelinated gaps called nodes of Ranvier (about 1 micrometer wide, every 1–2 mm). Action potentials regenerate at each node and the depolarization 'jumps' to the next node by passive current flow — saltatory conduction (Latin saltare = to leap). Conduction velocity in heavily myelinated A-alpha fibers reaches 80–120 m/s versus 0.5–2 m/s in unmyelinated C fibers of similar diameter. Multiple sclerosis demyelinates CNS axons, slowing or blocking conduction.
Why was the squid giant axon so important?
The squid giant axon, discovered by J. Z. Young in 1936, has a diameter of up to 1 mm — about 100 times larger than a typical mammalian axon — making it large enough to thread microelectrodes inside it. Alan Hodgkin and Andrew Huxley used this preparation between 1939 and 1952 to make the first direct intracellular recordings of action potentials, then to invent the voltage-clamp technique that holds membrane voltage fixed while recording the resulting current. The voltage-clamp let them dissect Na+ and K+ currents separately by changing extracellular ion concentrations and applying drugs (TEA, TTX, although the specific blockers came later). Their 1952 series in J. Physiol. produced the four-equation Hodgkin-Huxley model that quantitatively predicted action potential shape and propagation velocity from first principles. They shared the 1963 Nobel Prize for the work.